bibtex_url
stringlengths 41
50
| proceedings
stringlengths 38
47
| bibtext
stringlengths 709
3.56k
| abstract
stringlengths 17
2.11k
| authors
sequencelengths 1
72
| title
stringlengths 12
207
| id
stringlengths 7
16
| type
stringclasses 2
values | arxiv_id
stringlengths 0
10
| GitHub
sequencelengths 1
1
| paper_page
stringclasses 276
values | n_linked_authors
int64 -1
13
| upvotes
int64 -1
14
| num_comments
int64 -1
11
| n_authors
int64 -1
44
| paper_page_exists_pre_conf
int64 0
1
| Models
sequencelengths 0
100
| Datasets
sequencelengths 0
14
| Spaces
sequencelengths 0
100
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
https://aclanthology.org/2023.acl-long.900.bib | https://aclanthology.org/2023.acl-long.900/ | @inproceedings{liu-etal-2023-character,
title = "Character-Aware Models Improve Visual Text Rendering",
author = "Liu, Rosanne and
Garrette, Dan and
Saharia, Chitwan and
Chan, William and
Roberts, Adam and
Narang, Sharan and
Blok, Irina and
Mical, Rj and
Norouzi, Mohammad and
Constant, Noah",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.900",
doi = "10.18653/v1/2023.acl-long.900",
pages = "16270--16297",
abstract = "Current image generation models struggle to reliably produce well-formed visual text. In this paper, we investigate a key contributing factor: popular text-to-image models lack character-level input features, making it much harder to predict a word{'}s visual makeup as a series of glyphs. To quantify this effect, we conduct a series of experiments comparing character-aware vs. character-blind text encoders. In the text-only domain, we find that character-aware models provide large gains on a novel spelling task (WikiSpell). Applying our learnings to the visual domain, we train a suite of image generation models, and show that character-aware variants outperform their character-blind counterparts across a range of novel text rendering tasks (our DrawText benchmark). Our models set a much higher state-of-the-art on visual spelling, with 30+ point accuracy gains over competitors on rare words, despite training on far fewer examples.",
}
| Current image generation models struggle to reliably produce well-formed visual text. In this paper, we investigate a key contributing factor: popular text-to-image models lack character-level input features, making it much harder to predict a word{'}s visual makeup as a series of glyphs. To quantify this effect, we conduct a series of experiments comparing character-aware vs. character-blind text encoders. In the text-only domain, we find that character-aware models provide large gains on a novel spelling task (WikiSpell). Applying our learnings to the visual domain, we train a suite of image generation models, and show that character-aware variants outperform their character-blind counterparts across a range of novel text rendering tasks (our DrawText benchmark). Our models set a much higher state-of-the-art on visual spelling, with 30+ point accuracy gains over competitors on rare words, despite training on far fewer examples. | [
"Liu, Rosanne",
"Garrette, Dan",
"Saharia, Chitwan",
"Chan, William",
"Roberts, Adam",
"Narang, Sharan",
"Blok, Irina",
"Mical, Rj",
"Norouzi, Mohammad",
"Constant, Noah"
] | Character-Aware Models Improve Visual Text Rendering | acl-long.900 | Poster | 2212.10562 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.901.bib | https://aclanthology.org/2023.acl-long.901/ | @inproceedings{suwaileh-etal-2023-idrisi,
title = "{IDRISI}-{RA}: The First {A}rabic Location Mention Recognition Dataset of Disaster Tweets",
author = "Suwaileh, Reem and
Imran, Muhammad and
Elsayed, Tamer",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.901",
doi = "10.18653/v1/2023.acl-long.901",
pages = "16298--16317",
abstract = "Extracting geolocation information from social media data enables effective disaster management, as it helps response authorities; for example, in locating incidents for planning rescue activities, and affected people for evacuation. Nevertheless, geolocation extraction is greatly understudied for the low resource languages such as Arabic. To fill this gap, we introduce IDRISI-RA, the first publicly-available Arabic Location Mention Recognition (LMR) dataset that provides human- and automatically-labeled versions in order of thousands and millions of tweets, respectively. It contains both location mentions and their types (e.g., district, city). Our extensive analysis shows the decent geographical, domain, location granularity, temporal, and dialectical coverage of IDRISI-RA. Furthermore, we establish baselines using the standard Arabic NER models and build two simple, yet effective, LMR models. Our rigorous experiments confirm the need for developing specific models for Arabic LMR in the disaster domain. Moreover, experiments show the promising domain and geographical generalizability of IDRISI-RA under zero-shot learning.",
}
| Extracting geolocation information from social media data enables effective disaster management, as it helps response authorities; for example, in locating incidents for planning rescue activities, and affected people for evacuation. Nevertheless, geolocation extraction is greatly understudied for the low resource languages such as Arabic. To fill this gap, we introduce IDRISI-RA, the first publicly-available Arabic Location Mention Recognition (LMR) dataset that provides human- and automatically-labeled versions in order of thousands and millions of tweets, respectively. It contains both location mentions and their types (e.g., district, city). Our extensive analysis shows the decent geographical, domain, location granularity, temporal, and dialectical coverage of IDRISI-RA. Furthermore, we establish baselines using the standard Arabic NER models and build two simple, yet effective, LMR models. Our rigorous experiments confirm the need for developing specific models for Arabic LMR in the disaster domain. Moreover, experiments show the promising domain and geographical generalizability of IDRISI-RA under zero-shot learning. | [
"Suwaileh, Reem",
"Imran, Muhammad",
"Elsayed, Tamer"
] | IDRISI-RA: The First Arabic Location Mention Recognition Dataset of Disaster Tweets | acl-long.901 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.902.bib | https://aclanthology.org/2023.acl-long.902/ | @inproceedings{peng-etal-2023-fsuie,
title = "{FSUIE}: A Novel Fuzzy Span Mechanism for Universal Information Extraction",
author = "Peng, Tianshuo and
Li, Zuchao and
Zhang, Lefei and
Du, Bo and
Zhao, Hai",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.902",
doi = "10.18653/v1/2023.acl-long.902",
pages = "16318--16333",
abstract = "Universal Information Extraction (UIE) has been introduced as a unified framework for various Information Extraction (IE) tasks and has achieved widespread success. Despite this, UIE models have limitations. For example, they rely heavily on span boundaries in the data during training, which does not reflect the reality of span annotation challenges. Slight adjustments to positions can also meet requirements. Additionally, UIE models lack attention to the limited span length feature in IE. To address these deficiencies, we propose the Fuzzy Span Universal Information Extraction (FSUIE) framework. Specifically, our contribution consists of two concepts: \textit{fuzzy span loss} and \textit{fuzzy span attention}. Our experimental results on a series of main IE tasks show significant improvement compared to the baseline, especially in terms of fast convergence and strong performance with small amounts of data and training epochs. These results demonstrate the effectiveness and generalization of FSUIE in different tasks, settings, and scenarios.",
}
| Universal Information Extraction (UIE) has been introduced as a unified framework for various Information Extraction (IE) tasks and has achieved widespread success. Despite this, UIE models have limitations. For example, they rely heavily on span boundaries in the data during training, which does not reflect the reality of span annotation challenges. Slight adjustments to positions can also meet requirements. Additionally, UIE models lack attention to the limited span length feature in IE. To address these deficiencies, we propose the Fuzzy Span Universal Information Extraction (FSUIE) framework. Specifically, our contribution consists of two concepts: \textit{fuzzy span loss} and \textit{fuzzy span attention}. Our experimental results on a series of main IE tasks show significant improvement compared to the baseline, especially in terms of fast convergence and strong performance with small amounts of data and training epochs. These results demonstrate the effectiveness and generalization of FSUIE in different tasks, settings, and scenarios. | [
"Peng, Tianshuo",
"Li, Zuchao",
"Zhang, Lefei",
"Du, Bo",
"Zhao, Hai"
] | FSUIE: A Novel Fuzzy Span Mechanism for Universal Information Extraction | acl-long.902 | Poster | 2306.14913 | [
"https://github.com/pengts/fsuie"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.903.bib | https://aclanthology.org/2023.acl-long.903/ | @inproceedings{michael-etal-2023-nlp,
title = "What Do {NLP} Researchers Believe? Results of the {NLP} Community Metasurvey",
author = "Michael, Julian and
Holtzman, Ari and
Parrish, Alicia and
Mueller, Aaron and
Wang, Alex and
Chen, Angelica and
Madaan, Divyam and
Nangia, Nikita and
Pang, Richard Yuanzhe and
Phang, Jason and
Bowman, Samuel R.",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.903",
doi = "10.18653/v1/2023.acl-long.903",
pages = "16334--16368",
abstract = "We present the results of the NLP Community Metasurvey. Run from May to June 2022, it elicited opinions on controversial issues, including industry influence in the field, concerns about AGI, and ethics. Our results put concrete numbers to several controversies: For example, respondents are split in half on the importance of artificial general intelligence, whether language models understand language, and the necessity of linguistic structure and inductive bias for solving NLP problems. In addition, the survey posed meta-questions, asking respondents to predict the distribution of survey responses. This allows us to uncover false sociological beliefs where the community{'}s predictions don{'}t match reality. Among other results, we find that the community greatly overestimates its own belief in the usefulness of benchmarks and the potential for scaling to solve real-world problems, while underestimating its belief in the importance of linguistic structure, inductive bias, and interdisciplinary science.",
}
| We present the results of the NLP Community Metasurvey. Run from May to June 2022, it elicited opinions on controversial issues, including industry influence in the field, concerns about AGI, and ethics. Our results put concrete numbers to several controversies: For example, respondents are split in half on the importance of artificial general intelligence, whether language models understand language, and the necessity of linguistic structure and inductive bias for solving NLP problems. In addition, the survey posed meta-questions, asking respondents to predict the distribution of survey responses. This allows us to uncover false sociological beliefs where the community{'}s predictions don{'}t match reality. Among other results, we find that the community greatly overestimates its own belief in the usefulness of benchmarks and the potential for scaling to solve real-world problems, while underestimating its belief in the importance of linguistic structure, inductive bias, and interdisciplinary science. | [
"Michael, Julian",
"Holtzman, Ari",
"Parrish, Alicia",
"Mueller, Aaron",
"Wang, Alex",
"Chen, Angelica",
"Madaan, Divyam",
"Nangia, Nikita",
"Pang, Richard Yuanzhe",
"Phang, Jason",
"Bowman, Samuel R."
] | What Do NLP Researchers Believe? Results of the NLP Community Metasurvey | acl-long.903 | Poster | 2208.12852 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.904.bib | https://aclanthology.org/2023.acl-long.904/ | @inproceedings{yang-etal-2023-prototype,
title = "Prototype-Guided Pseudo Labeling for Semi-Supervised Text Classification",
author = "Yang, Weiyi and
Zhang, Richong and
Chen, Junfan and
Wang, Lihong and
Kim, Jaein",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.904",
doi = "10.18653/v1/2023.acl-long.904",
pages = "16369--16382",
abstract = "Semi-supervised text classification (SSTC) aims at text classification with few labeled data and massive unlabeled data. Recent works achieve this task by pseudo-labeling methods, with the belief that the unlabeled and labeled data have identical data distribution, and assign the unlabeled data with pseudo-labels as additional supervision. However, existing pseudo-labeling methods usually suffer from ambiguous categorical boundary issues when training the pseudo-labeling phase, and simply select pseudo-labels without considering the unbalanced categorical distribution of the unlabeled data, making it difficult to generate reliable pseudo-labels for each category. We propose a novel semi-supervised framework, namely ProtoS2, with prototypical cluster separation (PCS) and prototypical-center data selection (CDS) technology to address the issue. Particularly, PCS exploits categorical prototypes to assimilate instance representations within the same category, thus emphasizing low-density separation for the pseudo-labeled data to alleviate ambiguous boundaries. Besides, CDS selects central pseudo-labeled data considering the categorical distribution, avoiding the model from biasing on dominant categories. Empirical studies and extensive analysis with four benchmarks demonstrate the effectiveness of the proposed model.",
}
| Semi-supervised text classification (SSTC) aims at text classification with few labeled data and massive unlabeled data. Recent works achieve this task by pseudo-labeling methods, with the belief that the unlabeled and labeled data have identical data distribution, and assign the unlabeled data with pseudo-labels as additional supervision. However, existing pseudo-labeling methods usually suffer from ambiguous categorical boundary issues when training the pseudo-labeling phase, and simply select pseudo-labels without considering the unbalanced categorical distribution of the unlabeled data, making it difficult to generate reliable pseudo-labels for each category. We propose a novel semi-supervised framework, namely ProtoS2, with prototypical cluster separation (PCS) and prototypical-center data selection (CDS) technology to address the issue. Particularly, PCS exploits categorical prototypes to assimilate instance representations within the same category, thus emphasizing low-density separation for the pseudo-labeled data to alleviate ambiguous boundaries. Besides, CDS selects central pseudo-labeled data considering the categorical distribution, avoiding the model from biasing on dominant categories. Empirical studies and extensive analysis with four benchmarks demonstrate the effectiveness of the proposed model. | [
"Yang, Weiyi",
"Zhang, Richong",
"Chen, Junfan",
"Wang, Lihong",
"Kim, Jaein"
] | Prototype-Guided Pseudo Labeling for Semi-Supervised Text Classification | acl-long.904 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-long.905.bib | https://aclanthology.org/2023.acl-long.905/ | @inproceedings{maddela-etal-2023-lens,
title = "{LENS}: A Learnable Evaluation Metric for Text Simplification",
author = "Maddela, Mounica and
Dou, Yao and
Heineman, David and
Xu, Wei",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.905",
doi = "10.18653/v1/2023.acl-long.905",
pages = "16383--16408",
abstract = "Training learnable metrics using modern language models has recently emerged as a promising method for the automatic evaluation of machine translation. However, existing human evaluation datasets for text simplification have limited annotations that are based on unitary or outdated models, making them unsuitable for this approach. To address these issues, we introduce the SimpEval corpus that contains: SimpEval{\_}past, comprising 12K human ratings on 2.4K simplifications of 24 past systems, and SimpEval{\_}2022, a challenging simplification benchmark consisting of over 1K human ratings of 360 simplifications including GPT-3.5 generated text. Training on SimpEval, we present LENS, a Learnable Evaluation Metric for Text Simplification. Extensive empirical results show that LENS correlates much better with human judgment than existing metrics, paving the way for future progress in the evaluation of text simplification. We also introduce Rank {\&} Rate, a human evaluation framework that rates simplifications from several models in a list-wise manner using an interactive interface, which ensures both consistency and accuracy in the evaluation process and is used to create the SimpEval datasets.",
}
| Training learnable metrics using modern language models has recently emerged as a promising method for the automatic evaluation of machine translation. However, existing human evaluation datasets for text simplification have limited annotations that are based on unitary or outdated models, making them unsuitable for this approach. To address these issues, we introduce the SimpEval corpus that contains: SimpEval{\_}past, comprising 12K human ratings on 2.4K simplifications of 24 past systems, and SimpEval{\_}2022, a challenging simplification benchmark consisting of over 1K human ratings of 360 simplifications including GPT-3.5 generated text. Training on SimpEval, we present LENS, a Learnable Evaluation Metric for Text Simplification. Extensive empirical results show that LENS correlates much better with human judgment than existing metrics, paving the way for future progress in the evaluation of text simplification. We also introduce Rank {\&} Rate, a human evaluation framework that rates simplifications from several models in a list-wise manner using an interactive interface, which ensures both consistency and accuracy in the evaluation process and is used to create the SimpEval datasets. | [
"Maddela, Mounica",
"Dou, Yao",
"Heineman, David",
"Xu, Wei"
] | LENS: A Learnable Evaluation Metric for Text Simplification | acl-long.905 | Poster | 2212.09739 | [
"https://github.com/yao-dou/lens"
] | https://huggingface.co/papers/2212.09739 | 0 | 0 | 0 | 4 | 1 | [
"douy/T5-11B-Ctrl-Simplification"
] | [] | [] |
https://aclanthology.org/2023.acl-long.906.bib | https://aclanthology.org/2023.acl-long.906/ | @inproceedings{hu-etal-2023-meetingbank,
title = "{M}eeting{B}ank: A Benchmark Dataset for Meeting Summarization",
author = "Hu, Yebowen and
Ganter, Timothy and
Deilamsalehy, Hanieh and
Dernoncourt, Franck and
Foroosh, Hassan and
Liu, Fei",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.906",
doi = "10.18653/v1/2023.acl-long.906",
pages = "16409--16423",
abstract = "As the number of recorded meetings increases, it becomes increasingly important to utilize summarization technology to create useful summaries of these recordings. However, there is a crucial lack of annotated meeting corpora for developing this technology, as it can be hard to collect meetings, especially when the topics discussed are confidential. Furthermore, meeting summaries written by experienced writers are scarce, making it hard for abstractive summarizers to produce sensible output without a reliable reference. This lack of annotated corpora has hindered the development of meeting summarization technology. In this paper, we present MeetingBank, a new benchmark dataset of city council meetings over the past decade. MeetingBank is unique among other meeting corpora due to its divide-and-conquer approach, which involves dividing professionally written meeting minutes into shorter passages and aligning them with specific segments of the meeting. This breaks down the process of summarizing a lengthy meeting into smaller, more manageable tasks. The dataset provides a new testbed of various meeting summarization systems and also allows the public to gain insight into how council decisions are made. We make the collection, including meeting video links, transcripts, reference summaries, agenda, and other metadata, publicly available to facilitate the development of better meeting summarization techniques.",
}
| As the number of recorded meetings increases, it becomes increasingly important to utilize summarization technology to create useful summaries of these recordings. However, there is a crucial lack of annotated meeting corpora for developing this technology, as it can be hard to collect meetings, especially when the topics discussed are confidential. Furthermore, meeting summaries written by experienced writers are scarce, making it hard for abstractive summarizers to produce sensible output without a reliable reference. This lack of annotated corpora has hindered the development of meeting summarization technology. In this paper, we present MeetingBank, a new benchmark dataset of city council meetings over the past decade. MeetingBank is unique among other meeting corpora due to its divide-and-conquer approach, which involves dividing professionally written meeting minutes into shorter passages and aligning them with specific segments of the meeting. This breaks down the process of summarizing a lengthy meeting into smaller, more manageable tasks. The dataset provides a new testbed of various meeting summarization systems and also allows the public to gain insight into how council decisions are made. We make the collection, including meeting video links, transcripts, reference summaries, agenda, and other metadata, publicly available to facilitate the development of better meeting summarization techniques. | [
"Hu, Yebowen",
"Ganter, Timothy",
"Deilamsalehy, Hanieh",
"Dernoncourt, Franck",
"Foroosh, Hassan",
"Liu, Fei"
] | MeetingBank: A Benchmark Dataset for Meeting Summarization | acl-long.906 | Poster | 2305.17529 | [
""
] | https://huggingface.co/papers/2305.17529 | 1 | 0 | 0 | 6 | 1 | [] | [
"lytang/MeetingBank-transcript",
"huuuyeah/meetingbank",
"huuuyeah/MeetingBank_Audio",
"Aznor/MeetingBank-original"
] | [] |
https://aclanthology.org/2023.acl-long.907.bib | https://aclanthology.org/2023.acl-long.907/ | @inproceedings{ping-etal-2023-uniex,
title = "{U}ni{EX}: An Effective and Efficient Framework for Unified Information Extraction via a Span-extractive Perspective",
author = "Ping, Yang and
Lu, JunYu and
Gan, Ruyi and
Wang, Junjie and
Zhang, Yuxiang and
Zhang, Pingjian and
Zhang, Jiaxing",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.907",
doi = "10.18653/v1/2023.acl-long.907",
pages = "16424--16440",
abstract = "We propose a new paradigm for universal information extraction (IE) that is compatible with any schema format and applicable to a list of IE tasks, such as named entity recognition, relation extraction, event extraction and sentiment analysis. Our approach converts the text-based IE tasks as the token-pair problem, which uniformly disassembles all extraction targets into joint span detection, classification and association problems with a unified extractive framework, namely UniEX. UniEX can synchronously encode schema-based prompt and textual information, and collaboratively learn the generalized knowledge from pre-defined information using the auto-encoder language models. We develop a traffine attention mechanism to integrate heterogeneous factors including tasks, labels and inside tokens, and obtain the extraction target via a scoring matrix. Experiment results show that UniEX can outperform generative universal IE models in terms of performance and inference-speed on 14 benchmarks IE datasets with the supervised setting. The state-of-the-art performance in low-resource scenarios also verifies the transferability and effectiveness of UniEX.",
}
| We propose a new paradigm for universal information extraction (IE) that is compatible with any schema format and applicable to a list of IE tasks, such as named entity recognition, relation extraction, event extraction and sentiment analysis. Our approach converts the text-based IE tasks as the token-pair problem, which uniformly disassembles all extraction targets into joint span detection, classification and association problems with a unified extractive framework, namely UniEX. UniEX can synchronously encode schema-based prompt and textual information, and collaboratively learn the generalized knowledge from pre-defined information using the auto-encoder language models. We develop a traffine attention mechanism to integrate heterogeneous factors including tasks, labels and inside tokens, and obtain the extraction target via a scoring matrix. Experiment results show that UniEX can outperform generative universal IE models in terms of performance and inference-speed on 14 benchmarks IE datasets with the supervised setting. The state-of-the-art performance in low-resource scenarios also verifies the transferability and effectiveness of UniEX. | [
"Ping, Yang",
"Lu, JunYu",
"Gan, Ruyi",
"Wang, Junjie",
"Zhang, Yuxiang",
"Zhang, Pingjian",
"Zhang, Jiaxing"
] | UniEX: An Effective and Efficient Framework for Unified Information Extraction via a Span-extractive Perspective | acl-long.907 | Poster | 2305.10306 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-long.908.bib | https://aclanthology.org/2023.acl-long.908/ | @inproceedings{stodden-etal-2023-deplain,
title = "{DE}plain: A {G}erman Parallel Corpus with Intralingual Translations into Plain Language for Sentence and Document Simplification",
author = "Stodden, Regina and
Momen, Omar and
Kallmeyer, Laura",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.908",
doi = "10.18653/v1/2023.acl-long.908",
pages = "16441--16463",
abstract = "Text simplification is an intralingual translation task in which documents, or sentences of a complex source text are simplified for a target audience. The success of automatic text simplification systems is highly dependent on the quality of parallel data used for training and evaluation. To advance sentence simplification and document simplification in German, this paper presents DEplain, a new dataset of parallel, professionally written and manually aligned simplifications in plain German {``}plain DE{''} or in German: {``}Einfache Sprache{''}. DEplain consists of a news-domain (approx. 500 document pairs, approx. 13k sentence pairs) and a web-domain corpus (approx. 150 aligned documents, approx. 2k aligned sentence pairs). In addition, we are building a web harvester and experimenting with automatic alignment methods to facilitate the integration of non-aligned and to be-published parallel documents. Using this approach, we are dynamically increasing the web-domain corpus, so it is currently extended to approx. 750 document pairs and approx. 3.5k aligned sentence pairs. We show that using DEplain to train a transformer-based seq2seq text simplification model can achieve promising results. We make available the corpus, the adapted alignment methods for German, the web harvester and the trained models here: \url{https://github.com/rstodden/DEPlain}.",
}
| Text simplification is an intralingual translation task in which documents, or sentences of a complex source text are simplified for a target audience. The success of automatic text simplification systems is highly dependent on the quality of parallel data used for training and evaluation. To advance sentence simplification and document simplification in German, this paper presents DEplain, a new dataset of parallel, professionally written and manually aligned simplifications in plain German {``}plain DE{''} or in German: {``}Einfache Sprache{''}. DEplain consists of a news-domain (approx. 500 document pairs, approx. 13k sentence pairs) and a web-domain corpus (approx. 150 aligned documents, approx. 2k aligned sentence pairs). In addition, we are building a web harvester and experimenting with automatic alignment methods to facilitate the integration of non-aligned and to be-published parallel documents. Using this approach, we are dynamically increasing the web-domain corpus, so it is currently extended to approx. 750 document pairs and approx. 3.5k aligned sentence pairs. We show that using DEplain to train a transformer-based seq2seq text simplification model can achieve promising results. We make available the corpus, the adapted alignment methods for German, the web harvester and the trained models here: \url{https://github.com/rstodden/DEPlain}. | [
"Stodden, Regina",
"Momen, Omar",
"Kallmeyer, Laura"
] | DEplain: A German Parallel Corpus with Intralingual Translations into Plain Language for Sentence and Document Simplification | acl-long.908 | Poster | 2305.18939 | [
"https://github.com/rstodden/deplain"
] | https://huggingface.co/papers/2305.18939 | 2 | 0 | 0 | 3 | 1 | [
"DEplain/trimmed_mbart_sents_apa_web",
"DEplain/trimmed_longmbart_docs_apa",
"DEplain/trimmed_mbart_sents_apa"
] | [
"DEplain/DEplain-web-doc",
"DEplain/DEplain-APA-sent",
"DEplain/DEplain-APA-doc",
"DEplain/DEplain-web-sent"
] | [] |
https://aclanthology.org/2023.acl-long.909.bib | https://aclanthology.org/2023.acl-long.909/ | @inproceedings{li-etal-2023-neural,
title = "A Neural Divide-and-Conquer Reasoning Framework for Image Retrieval from Linguistically Complex Text",
author = "Li, Yunxin and
Hu, Baotian and
Ding, Yuxin and
Ma, Lin and
Zhang, Min",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.909",
doi = "10.18653/v1/2023.acl-long.909",
pages = "16464--16476",
abstract = "Pretrained Vision-Language Models (VLMs) have achieved remarkable performance in image retrieval from text. However, their performance drops drastically when confronted with linguistically complex texts that they struggle to comprehend. Inspired by the Divide-and-Conquer algorithm and dual-process theory, in this paper, we regard linguistically complex texts as compound proposition texts composed of multiple simple proposition sentences and propose an end-to-end Neural Divide-and-Conquer Reasoning framework, dubbed NDCR. It contains three main components: 1) Divide: a proposition generator divides the compound proposition text into simple proposition sentences and produces their corresponding representations, 2) Conquer: a pretrained VLMs-based visual-linguistic interactor achieves the interaction between decomposed proposition sentences and images, 3) Combine: a neural-symbolic reasoner combines the above reasoning states to obtain the final solution via a neural logic reasoning approach. According to the dual-process theory, the visual-linguistic interactor and neural-symbolic reasoner could be regarded as analogical reasoning System 1 and logical reasoning System 2. We conduct extensive experiments on a challenging image retrieval from contextual descriptions data set. Experimental results and analyses indicate NDCR significantly improves performance in the complex image-text reasoning problem.",
}
| Pretrained Vision-Language Models (VLMs) have achieved remarkable performance in image retrieval from text. However, their performance drops drastically when confronted with linguistically complex texts that they struggle to comprehend. Inspired by the Divide-and-Conquer algorithm and dual-process theory, in this paper, we regard linguistically complex texts as compound proposition texts composed of multiple simple proposition sentences and propose an end-to-end Neural Divide-and-Conquer Reasoning framework, dubbed NDCR. It contains three main components: 1) Divide: a proposition generator divides the compound proposition text into simple proposition sentences and produces their corresponding representations, 2) Conquer: a pretrained VLMs-based visual-linguistic interactor achieves the interaction between decomposed proposition sentences and images, 3) Combine: a neural-symbolic reasoner combines the above reasoning states to obtain the final solution via a neural logic reasoning approach. According to the dual-process theory, the visual-linguistic interactor and neural-symbolic reasoner could be regarded as analogical reasoning System 1 and logical reasoning System 2. We conduct extensive experiments on a challenging image retrieval from contextual descriptions data set. Experimental results and analyses indicate NDCR significantly improves performance in the complex image-text reasoning problem. | [
"Li, Yunxin",
"Hu, Baotian",
"Ding, Yuxin",
"Ma, Lin",
"Zhang, Min"
] | A Neural Divide-and-Conquer Reasoning Framework for Image Retrieval from Linguistically Complex Text | acl-long.909 | Poster | 2305.02265 | [
"https://github.com/yunxinli/ndcr"
] | https://huggingface.co/papers/2305.02265 | 1 | 0 | 0 | 5 | 1 | [
"YunxinLi/pretrain_BART_generator_coldstart_OFA"
] | [] | [] |
https://aclanthology.org/2023.acl-long.910.bib | https://aclanthology.org/2023.acl-long.910/ | @inproceedings{gao-etal-2023-rarr,
title = "{RARR}: Researching and Revising What Language Models Say, Using Language Models",
author = "Gao, Luyu and
Dai, Zhuyun and
Pasupat, Panupong and
Chen, Anthony and
Chaganty, Arun Tejasvi and
Fan, Yicheng and
Zhao, Vincent and
Lao, Ni and
Lee, Hongrae and
Juan, Da-Cheng and
Guu, Kelvin",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-long.910",
doi = "10.18653/v1/2023.acl-long.910",
pages = "16477--16508",
abstract = "Language models (LMs) now excel at many tasks such as question answering, reasoning, and dialog. However, they sometimes generate unsupported or misleading content. A user cannot easily determine whether their outputs are trustworthy or not, because most LMs do not have any built-in mechanism for attribution to external evidence. To enable attribution while still preserving all the powerful advantages of recent generation models, we propose RARR (Retrofit Attribution using Research and Revision), a system that 1) automatically finds attribution for the output of any text generation model, and 2) post-edits the output to fix unsupported content while preserving the original output as much as possible. When applied to the output of several state-of-the-art LMs on a diverse set of generation tasks, we find that RARR significantly improves attribution while otherwise preserving the original input to a much greater degree than previously explored edit models. Furthermore, the implementation of RARR requires only a handful of training examples, a large language model, and standard web search.",
}
| Language models (LMs) now excel at many tasks such as question answering, reasoning, and dialog. However, they sometimes generate unsupported or misleading content. A user cannot easily determine whether their outputs are trustworthy or not, because most LMs do not have any built-in mechanism for attribution to external evidence. To enable attribution while still preserving all the powerful advantages of recent generation models, we propose RARR (Retrofit Attribution using Research and Revision), a system that 1) automatically finds attribution for the output of any text generation model, and 2) post-edits the output to fix unsupported content while preserving the original output as much as possible. When applied to the output of several state-of-the-art LMs on a diverse set of generation tasks, we find that RARR significantly improves attribution while otherwise preserving the original input to a much greater degree than previously explored edit models. Furthermore, the implementation of RARR requires only a handful of training examples, a large language model, and standard web search. | [
"Gao, Luyu",
"Dai, Zhuyun",
"Pasupat, Panupong",
"Chen, Anthony",
"Chaganty, Arun Tejasvi",
"Fan, Yicheng",
"Zhao, Vincent",
"Lao, Ni",
"Lee, Hongrae",
"Juan, Da-Cheng",
"Guu, Kelvin"
] | RARR: Researching and Revising What Language Models Say, Using Language Models | acl-long.910 | Poster | 2210.08726 | [
"https://github.com/anthonywchen/rarr"
] | https://huggingface.co/papers/2210.08726 | 0 | 1 | 2 | 11 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-short.1.bib | https://aclanthology.org/2023.acl-short.1/ | @inproceedings{chirkova-etal-2023-marginalize,
title = "Should you marginalize over possible tokenizations?",
author = "Chirkova, Nadezhda and
Kruszewski, Germ{\'a}n and
Rozen, Jos and
Dymetman, Marc",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.1",
doi = "10.18653/v1/2023.acl-short.1",
pages = "1--12",
abstract = "Autoregressive language models (LMs) map token sequences to probabilities. The usual practice for computing the probability of any character string (e.g. English sentences) is to first transform it into a sequence of tokens that is scored by the model. However, there are exponentially many token sequences that represent any given string. To truly compute the probability of a string one should marginalize over all tokenizations, which is typically intractable. Here, we analyze whether the practice of ignoring the marginalization is justified. To this end, we devise an importance-sampling-based algorithm that allows us to compute estimates of the marginal probabilities and compare them to the default procedure in a range of state-of-the-art models and datasets. Our results show that the gap in log-likelihood is no larger than 0.5{\%} in most cases, but that it becomes more pronounced for data with long complex words.",
}
| Autoregressive language models (LMs) map token sequences to probabilities. The usual practice for computing the probability of any character string (e.g. English sentences) is to first transform it into a sequence of tokens that is scored by the model. However, there are exponentially many token sequences that represent any given string. To truly compute the probability of a string one should marginalize over all tokenizations, which is typically intractable. Here, we analyze whether the practice of ignoring the marginalization is justified. To this end, we devise an importance-sampling-based algorithm that allows us to compute estimates of the marginal probabilities and compare them to the default procedure in a range of state-of-the-art models and datasets. Our results show that the gap in log-likelihood is no larger than 0.5{\%} in most cases, but that it becomes more pronounced for data with long complex words. | [
"Chirkova, Nadezhda",
"Kruszewski, Germ{\\'a}n",
"Rozen, Jos",
"Dymetman, Marc"
] | Should you marginalize over possible tokenizations? | acl-short.1 | Poster | 2306.17757 | [
"https://github.com/naver/marginalization"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.2.bib | https://aclanthology.org/2023.acl-short.2/ | @inproceedings{yoshinaga-2023-back,
title = "Back to Patterns: Efficient {J}apanese Morphological Analysis with Feature-Sequence Trie",
author = "Yoshinaga, Naoki",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.2",
doi = "10.18653/v1/2023.acl-short.2",
pages = "13--23",
abstract = "Accurate neural models are much less efficient than non-neural models and are useless for processing billions of social media posts or handling user queries in real time with a limited budget. This study revisits the fastest pattern-based NLP methods to make them as accurate as possible, thus yielding a strikingly simple yet surprisingly accurate morphological analyzer for Japanese. The proposed method induces reliable patterns from a morphological dictionary and annotated data. Experimental results on two standard datasets confirm that the method exhibits comparable accuracy to learning-based baselines, while boasting a remarkable throughput of over 1,000,000 sentences per second on a single modern CPU. The source code is available at \url{https://www.tkl.iis.u-tokyo.ac.jp/ynaga/jagger/}",
}
| Accurate neural models are much less efficient than non-neural models and are useless for processing billions of social media posts or handling user queries in real time with a limited budget. This study revisits the fastest pattern-based NLP methods to make them as accurate as possible, thus yielding a strikingly simple yet surprisingly accurate morphological analyzer for Japanese. The proposed method induces reliable patterns from a morphological dictionary and annotated data. Experimental results on two standard datasets confirm that the method exhibits comparable accuracy to learning-based baselines, while boasting a remarkable throughput of over 1,000,000 sentences per second on a single modern CPU. The source code is available at \url{https://www.tkl.iis.u-tokyo.ac.jp/ynaga/jagger/} | [
"Yoshinaga, Naoki"
] | Back to Patterns: Efficient Japanese Morphological Analysis with Feature-Sequence Trie | acl-short.2 | Poster | 2305.19045 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.3.bib | https://aclanthology.org/2023.acl-short.3/ | @inproceedings{kim-etal-2023-transformed,
title = "Transformed Protoform Reconstruction",
author = "Kim, Young Min and
Chang, Kalvin and
Cui, Chenxuan and
Mortensen, David R.",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.3",
doi = "10.18653/v1/2023.acl-short.3",
pages = "24--38",
abstract = "Protoform reconstruction is the task of inferring what morphemes or words appeared like in the ancestral languages of a set of daughter languages. Meloni et al (2021) achieved the state-of-the-art on Latin protoform reconstruction with an RNN-based encoder-decoder with attention model. We update their model with the state-of-the-art seq2seq model: the Transformer. Our model outperforms their model on a suite of different metrics on two different datasets: their Romance data of 8,000 cognates spanning 5 languages and a Chinese dataset (Hou 2004) of 800+ cognates spanning 39 varieties. We also probe our model for potential phylogenetic signal contained in the model. Our code is publicly available at \url{https://github.com/cmu-llab/acl-2023}.",
}
| Protoform reconstruction is the task of inferring what morphemes or words appeared like in the ancestral languages of a set of daughter languages. Meloni et al (2021) achieved the state-of-the-art on Latin protoform reconstruction with an RNN-based encoder-decoder with attention model. We update their model with the state-of-the-art seq2seq model: the Transformer. Our model outperforms their model on a suite of different metrics on two different datasets: their Romance data of 8,000 cognates spanning 5 languages and a Chinese dataset (Hou 2004) of 800+ cognates spanning 39 varieties. We also probe our model for potential phylogenetic signal contained in the model. Our code is publicly available at \url{https://github.com/cmu-llab/acl-2023}. | [
"Kim, Young Min",
"Chang, Kalvin",
"Cui, Chenxuan",
"Mortensen, David R."
] | Transformed Protoform Reconstruction | acl-short.3 | Oral | 2307.01896 | [
"https://github.com/cmu-llab/acl-2023"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.4.bib | https://aclanthology.org/2023.acl-short.4/ | @inproceedings{hardt-2023-ellipsis,
title = "Ellipsis-Dependent Reasoning: a New Challenge for Large Language Models",
author = "Hardt, Daniel",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.4",
doi = "10.18653/v1/2023.acl-short.4",
pages = "39--47",
abstract = "We propose a novel challenge for large language models: ellipsis-dependent reasoning. We define several structures of paired examples, where an ellipsis example is matched to its non-ellipsis counterpart, and a question is posed which requires resolution of the ellipsis. Test results show that the best models perform well on non-elliptical examples but struggle with all but the simplest ellipsis structures.",
}
| We propose a novel challenge for large language models: ellipsis-dependent reasoning. We define several structures of paired examples, where an ellipsis example is matched to its non-ellipsis counterpart, and a question is posed which requires resolution of the ellipsis. Test results show that the best models perform well on non-elliptical examples but struggle with all but the simplest ellipsis structures. | [
"Hardt, Daniel"
] | Ellipsis-Dependent Reasoning: a New Challenge for Large Language Models | acl-short.4 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.5.bib | https://aclanthology.org/2023.acl-short.5/ | @inproceedings{tang-surdeanu-2023-bootstrapping,
title = "Bootstrapping Neural Relation and Explanation Classifiers",
author = "Tang, Zheng and
Surdeanu, Mihai",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.5",
doi = "10.18653/v1/2023.acl-short.5",
pages = "48--56",
abstract = "We introduce a method that self trains (or bootstraps) neural relation and explanation classifiers. Our work expands the supervised approach of CITATION, which jointly trains a relation classifier with an explanation classifier that identifies context words important for the relation at hand, to semi-supervised scenarios. In particular, our approach iteratively converts the explainable models{'} outputs to rules and applies them to unlabeled text to produce new annotations. Our evaluation on the TACRED dataset shows that our method outperforms the rule-based model we started from by 15 F1 points, outperforms traditional self-training that relies just on the relation classifier by 5 F1 points, and performs comparatively with the prompt-based approach of CITATION (without requiring an additional natural language inference component).",
}
| We introduce a method that self trains (or bootstraps) neural relation and explanation classifiers. Our work expands the supervised approach of CITATION, which jointly trains a relation classifier with an explanation classifier that identifies context words important for the relation at hand, to semi-supervised scenarios. In particular, our approach iteratively converts the explainable models{'} outputs to rules and applies them to unlabeled text to produce new annotations. Our evaluation on the TACRED dataset shows that our method outperforms the rule-based model we started from by 15 F1 points, outperforms traditional self-training that relies just on the relation classifier by 5 F1 points, and performs comparatively with the prompt-based approach of CITATION (without requiring an additional natural language inference component). | [
"Tang, Zheng",
"Surdeanu, Mihai"
] | Bootstrapping Neural Relation and Explanation Classifiers | acl-short.5 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.6.bib | https://aclanthology.org/2023.acl-short.6/ | @inproceedings{nowak-cotterell-2023-fast,
title = "A Fast Algorithm for Computing Prefix Probabilities",
author = "Nowak, Franz and
Cotterell, Ryan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.6",
doi = "10.18653/v1/2023.acl-short.6",
pages = "57--69",
abstract = "Multiple algorithms are known for efficiently calculating the prefix probability of a string under a probabilistic context-free grammar (PCFG). Good algorithms for the problem have a runtime cubic in the length of the input string. However, some proposed algorithms are suboptimal with respect to the size of the grammar. This paper proposes a new speed-up of Jelinek and Lafferty{'}s (1991) algorithm, which runs in $O(n^3|N|^3 + |N|^4)$, where n is the input length and |N| is the number of non-terminals in the grammar. In contrast, our speed-up runs in $O(n^2|N|^3 + n^3|N|^2)$.",
}
| Multiple algorithms are known for efficiently calculating the prefix probability of a string under a probabilistic context-free grammar (PCFG). Good algorithms for the problem have a runtime cubic in the length of the input string. However, some proposed algorithms are suboptimal with respect to the size of the grammar. This paper proposes a new speed-up of Jelinek and Lafferty{'}s (1991) algorithm, which runs in $O(n^3|N|^3 + |N|^4)$, where n is the input length and |N| is the number of non-terminals in the grammar. In contrast, our speed-up runs in $O(n^2|N|^3 + n^3|N|^2)$. | [
"Nowak, Franz",
"Cotterell, Ryan"
] | A Fast Algorithm for Computing Prefix Probabilities | acl-short.6 | Poster | 2306.02303 | [
"https://github.com/rycolab/prefix-parsing"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.7.bib | https://aclanthology.org/2023.acl-short.7/ | @inproceedings{gonzalez-gutierrez-etal-2023-analyzing,
title = "Analyzing Text Representations by Measuring Task Alignment",
author = "Gonzalez-Gutierrez, Cesar and
Primadhanty, Audi and
Cazzaro, Francesco and
Quattoni, Ariadna",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.7",
doi = "10.18653/v1/2023.acl-short.7",
pages = "70--81",
abstract = "Textual representations based on pre-trained language models are key, especially in few-shot learning scenarios. What makes a representation good for text classification? Is it due to the geometric properties of the space or because it is well aligned with the task? We hypothesize the second claim. To test it, we develop a task alignment score based on hierarchical clustering that measures alignment at different levels of granularity. Our experiments on text classification validate our hypothesis by showing that task alignment can explain the classification performance of a given representation.",
}
| Textual representations based on pre-trained language models are key, especially in few-shot learning scenarios. What makes a representation good for text classification? Is it due to the geometric properties of the space or because it is well aligned with the task? We hypothesize the second claim. To test it, we develop a task alignment score based on hierarchical clustering that measures alignment at different levels of granularity. Our experiments on text classification validate our hypothesis by showing that task alignment can explain the classification performance of a given representation. | [
"Gonzalez-Gutierrez, Cesar",
"Primadhanty, Audi",
"Cazzaro, Francesco",
"Quattoni, Ariadna"
] | Analyzing Text Representations by Measuring Task Alignment | acl-short.7 | Poster | 2305.19747 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.8.bib | https://aclanthology.org/2023.acl-short.8/ | @inproceedings{khare-etal-2023-tracing,
title = "Tracing Linguistic Markers of Influence in a Large Online Organisation",
author = "Khare, Prashant and
Shekhar, Ravi and
Karan, Mladen and
McQuistin, Stephen and
Perkins, Colin and
Castro, Ignacio and
Tyson, Gareth and
Healey, Patrick and
Purver, Matthew",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.8",
doi = "10.18653/v1/2023.acl-short.8",
pages = "82--90",
abstract = "Social science and psycholinguistic research have shown that power and status affect how people use language in a range of domains. Here, we investigate a similar question in a large, distributed, consensus-driven community with little traditional power hierarchy {--} the Internet Engineering Task Force (IETF), a collaborative organisation that designs internet standards. Our analysis based on lexical categories (LIWC) and BERT, shows that participants{'} levels of influence can be predicted from their email text, and identify key linguistic differences (e.g., certain LIWC categories, such as {``}WE{''} are positively correlated with high-influence). We also identify the differences in language use for the same person before and after becoming influential.",
}
| Social science and psycholinguistic research have shown that power and status affect how people use language in a range of domains. Here, we investigate a similar question in a large, distributed, consensus-driven community with little traditional power hierarchy {--} the Internet Engineering Task Force (IETF), a collaborative organisation that designs internet standards. Our analysis based on lexical categories (LIWC) and BERT, shows that participants{'} levels of influence can be predicted from their email text, and identify key linguistic differences (e.g., certain LIWC categories, such as {``}WE{''} are positively correlated with high-influence). We also identify the differences in language use for the same person before and after becoming influential. | [
"Khare, Prashant",
"Shekhar, Ravi",
"Karan, Mladen",
"McQuistin, Stephen",
"Perkins, Colin",
"Castro, Ignacio",
"Tyson, Gareth",
"Healey, Patrick",
"Purver, Matthew"
] | Tracing Linguistic Markers of Influence in a Large Online Organisation | acl-short.8 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.9.bib | https://aclanthology.org/2023.acl-short.9/ | @inproceedings{li-etal-2023-metaphor,
title = "Metaphor Detection via Explicit Basic Meanings Modelling",
author = "Li, Yucheng and
Wang, Shun and
Lin, Chenghua and
Guerin, Frank",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.9",
doi = "10.18653/v1/2023.acl-short.9",
pages = "91--100",
abstract = "One noticeable trend in metaphor detection is the embrace of linguistic theories such as the metaphor identification procedure (MIP) for model architecture design. While MIP clearly defines that the metaphoricity of a lexical unit is determined based on the contrast between its contextual meaning and its basic meaning, existing work does not strictly follow this principle, typically using the aggregated meaning to approximate the basic meaning of target words. In this paper, we propose a novel metaphor detection method, which models the basic meaning of the word based on literal annotation from the training set, and then compares this with the contextual meaning in a target sentence to identify metaphors. Empirical results show that our method outperforms the state-of-the-art method significantly by 1.0{\%} in F1 score. Moreover, our performance even reaches the theoretical upper bound on the VUA18 benchmark for targets with basic annotations, which demonstrates the importance of modelling basic meanings for metaphor detection.",
}
| One noticeable trend in metaphor detection is the embrace of linguistic theories such as the metaphor identification procedure (MIP) for model architecture design. While MIP clearly defines that the metaphoricity of a lexical unit is determined based on the contrast between its contextual meaning and its basic meaning, existing work does not strictly follow this principle, typically using the aggregated meaning to approximate the basic meaning of target words. In this paper, we propose a novel metaphor detection method, which models the basic meaning of the word based on literal annotation from the training set, and then compares this with the contextual meaning in a target sentence to identify metaphors. Empirical results show that our method outperforms the state-of-the-art method significantly by 1.0{\%} in F1 score. Moreover, our performance even reaches the theoretical upper bound on the VUA18 benchmark for targets with basic annotations, which demonstrates the importance of modelling basic meanings for metaphor detection. | [
"Li, Yucheng",
"Wang, Shun",
"Lin, Chenghua",
"Guerin, Frank"
] | Metaphor Detection via Explicit Basic Meanings Modelling | acl-short.9 | Poster | 2305.17268 | [
"https://github.com/liyucheng09/basicbert"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.10.bib | https://aclanthology.org/2023.acl-short.10/ | @inproceedings{chen-etal-2023-xsim,
title = "x{SIM}++: An Improved Proxy to Bitext Mining Performance for Low-Resource Languages",
author = "Chen, Mingda and
Heffernan, Kevin and
{\c{C}}elebi, Onur and
Mourachko, Alexandre and
Schwenk, Holger",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.10",
doi = "10.18653/v1/2023.acl-short.10",
pages = "101--109",
abstract = "We introduce a new proxy score for evaluating bitext mining based on similarity in a multilingual embedding space: xsim++. In comparison to xsim, this improved proxy leverages rule-based approaches to extend English sentences in any evaluation set with synthetic, hard-to-distinguish examples which more closely mirror the scenarios we encounter during large-scale mining. We validate this proxy by running a significant number of bitext mining experiments for a set of low-resource languages, and subsequently train NMT systems on the mined data. In comparison to xsim, we show that xsim++ is better correlated with the downstream BLEU scores of translation systems trained on mined bitexts, providing a reliable proxy of bitext mining performance without needing to run expensive bitext mining pipelines. xsim++ also reports performance for different error types, offering more fine-grained feedbacks for model development.",
}
| We introduce a new proxy score for evaluating bitext mining based on similarity in a multilingual embedding space: xsim++. In comparison to xsim, this improved proxy leverages rule-based approaches to extend English sentences in any evaluation set with synthetic, hard-to-distinguish examples which more closely mirror the scenarios we encounter during large-scale mining. We validate this proxy by running a significant number of bitext mining experiments for a set of low-resource languages, and subsequently train NMT systems on the mined data. In comparison to xsim, we show that xsim++ is better correlated with the downstream BLEU scores of translation systems trained on mined bitexts, providing a reliable proxy of bitext mining performance without needing to run expensive bitext mining pipelines. xsim++ also reports performance for different error types, offering more fine-grained feedbacks for model development. | [
"Chen, Mingda",
"Heffernan, Kevin",
"{\\c{C}}elebi, Onur",
"Mourachko, Alex",
"re",
"Schwenk, Holger"
] | xSIM++: An Improved Proxy to Bitext Mining Performance for Low-Resource Languages | acl-short.10 | Oral | 2306.12907 | [
"https://github.com/facebookresearch/LASER"
] | https://huggingface.co/papers/2306.12907 | 0 | 0 | 0 | 5 | 1 | [] | [
"jaygala24/xsimplusplus"
] | [] |
https://aclanthology.org/2023.acl-short.11.bib | https://aclanthology.org/2023.acl-short.11/ | @inproceedings{cai-etal-2023-graph,
title = "Improving Low-resource Named Entity Recognition with Graph Propagated Data Augmentation",
author = "Cai, Jiong and
Huang, Shen and
Jiang, Yong and
Tan, Zeqi and
Xie, Pengjun and
Tu, Kewei",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.11",
doi = "10.18653/v1/2023.acl-short.11",
pages = "110--118",
abstract = "Data augmentation is an effective solution to improve model performance and robustness for low-resource named entity recognition (NER). However, synthetic data often suffer from poor diversity, which leads to performance limitations. In this paper, we propose a novel Graph Propagated Data Augmentation (GPDA) framework for Named Entity Recognition (NER), leveraging graph propagation to build relationships between labeled data and unlabeled natural texts. By projecting the annotations from the labeled text to the unlabeled text, the unlabeled texts are partially labeled, which has more diversity rather than synthetic annotated data. To strengthen the propagation precision, a simple search engine built on Wikipedia is utilized to fetch related texts of labeled data and to propagate the entity labels to them in the light of the anchor links. Besides, we construct and perform experiments on a real-world low-resource dataset of the E-commerce domain, which will be publicly available to facilitate the low-resource NER research. Experimental results show that GPDA presents substantial improvements over previous data augmentation methods on multiple low-resource NER datasets.",
}
| Data augmentation is an effective solution to improve model performance and robustness for low-resource named entity recognition (NER). However, synthetic data often suffer from poor diversity, which leads to performance limitations. In this paper, we propose a novel Graph Propagated Data Augmentation (GPDA) framework for Named Entity Recognition (NER), leveraging graph propagation to build relationships between labeled data and unlabeled natural texts. By projecting the annotations from the labeled text to the unlabeled text, the unlabeled texts are partially labeled, which has more diversity rather than synthetic annotated data. To strengthen the propagation precision, a simple search engine built on Wikipedia is utilized to fetch related texts of labeled data and to propagate the entity labels to them in the light of the anchor links. Besides, we construct and perform experiments on a real-world low-resource dataset of the E-commerce domain, which will be publicly available to facilitate the low-resource NER research. Experimental results show that GPDA presents substantial improvements over previous data augmentation methods on multiple low-resource NER datasets. | [
"Cai, Jiong",
"Huang, Shen",
"Jiang, Yong",
"Tan, Zeqi",
"Xie, Pengjun",
"Tu, Kewei"
] | Improving Low-resource Named Entity Recognition with Graph Propagated Data Augmentation | acl-short.11 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.12.bib | https://aclanthology.org/2023.acl-short.12/ | @inproceedings{maekawa-etal-2023-dataset,
title = "Dataset Distillation with Attention Labels for Fine-tuning {BERT}",
author = "Maekawa, Aru and
Kobayashi, Naoki and
Funakoshi, Kotaro and
Okumura, Manabu",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.12",
doi = "10.18653/v1/2023.acl-short.12",
pages = "119--127",
abstract = "Dataset distillation aims to create a small dataset of informative synthetic samples to rapidly train neural networks that retain the performance of the original dataset. In this paper, we focus on constructing distilled few-shot datasets for natural language processing (NLP) tasks to fine-tune pre-trained transformers. Specifically, we propose to introduce attention labels, which can efficiently distill the knowledge from the original dataset and transfer it to the transformer models via attention probabilities. We evaluated our dataset distillation methods in four various NLP tasks and demonstrated that it is possible to create distilled few-shot datasets with the attention labels, yielding impressive performances for fine-tuning BERT. Specifically, in AGNews, a four-class news classification task, our distilled few-shot dataset achieved up to 93.2{\%} accuracy, which is 98.5{\%} performance of the original dataset even with only one sample per class and only one gradient step.",
}
| Dataset distillation aims to create a small dataset of informative synthetic samples to rapidly train neural networks that retain the performance of the original dataset. In this paper, we focus on constructing distilled few-shot datasets for natural language processing (NLP) tasks to fine-tune pre-trained transformers. Specifically, we propose to introduce attention labels, which can efficiently distill the knowledge from the original dataset and transfer it to the transformer models via attention probabilities. We evaluated our dataset distillation methods in four various NLP tasks and demonstrated that it is possible to create distilled few-shot datasets with the attention labels, yielding impressive performances for fine-tuning BERT. Specifically, in AGNews, a four-class news classification task, our distilled few-shot dataset achieved up to 93.2{\%} accuracy, which is 98.5{\%} performance of the original dataset even with only one sample per class and only one gradient step. | [
"Maekawa, Aru",
"Kobayashi, Naoki",
"Funakoshi, Kotaro",
"Okumura, Manabu"
] | Dataset Distillation with Attention Labels for Fine-tuning BERT | acl-short.12 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.13.bib | https://aclanthology.org/2023.acl-short.13/ | @inproceedings{puduppully-etal-2023-multi,
title = "Multi-Document Summarization with Centroid-Based Pretraining",
author = "Puduppully, Ratish Surendran and
Jain, Parag and
Chen, Nancy and
Steedman, Mark",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.13",
doi = "10.18653/v1/2023.acl-short.13",
pages = "128--138",
abstract = "In Multi-Document Summarization (MDS), the input can be modeled as a set of documents, and the output is its summary. In this paper, we focus on pretraining objectives for MDS. Specifically, we introduce a novel pretraining objective, which involves selecting the ROUGE-based centroid of each document cluster as a proxy for its summary. Our objective thus does not require human written summaries and can be utilized for pretraining on a dataset consisting solely of document sets. Through zero-shot, few-shot, and fully supervised experiments on multiple MDS datasets, we show that our model \textit{Centrum} is better or comparable to a state-of-the-art model. We make the pretrained and fine-tuned models freely available to the research community \url{https://github.com/ratishsp/centrum}.",
}
| In Multi-Document Summarization (MDS), the input can be modeled as a set of documents, and the output is its summary. In this paper, we focus on pretraining objectives for MDS. Specifically, we introduce a novel pretraining objective, which involves selecting the ROUGE-based centroid of each document cluster as a proxy for its summary. Our objective thus does not require human written summaries and can be utilized for pretraining on a dataset consisting solely of document sets. Through zero-shot, few-shot, and fully supervised experiments on multiple MDS datasets, we show that our model \textit{Centrum} is better or comparable to a state-of-the-art model. We make the pretrained and fine-tuned models freely available to the research community \url{https://github.com/ratishsp/centrum}. | [
"Puduppully, Ratish Surendran",
"Jain, Parag",
"Chen, Nancy",
"Steedman, Mark"
] | Multi-Document Summarization with Centroid-Based Pretraining | acl-short.13 | Poster | 2208.01006 | [
"https://github.com/ratishsp/centrum"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.14.bib | https://aclanthology.org/2023.acl-short.14/ | @inproceedings{de-varda-marelli-2023-scaling,
title = "Scaling in Cognitive Modelling: a Multilingual Approach to Human Reading Times",
author = "de Varda, Andrea and
Marelli, Marco",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.14",
doi = "10.18653/v1/2023.acl-short.14",
pages = "139--149",
abstract = "Neural language models are increasingly valued in computational psycholinguistics, due to their ability to provide conditional probability distributions over the lexicon that are predictive of human processing times. Given the vast array of available models, it is of both theoretical and methodological importance to assess what features of a model influence its psychometric quality. In this work we focus on parameter size, showing that larger Transformer-based language models generate probabilistic estimates that are less predictive of early eye-tracking measurements reflecting lexical access and early semantic integration. However, relatively bigger models show an advantage in capturing late eye-tracking measurements that reflect the full semantic and syntactic integration of a word into the current language context. Our results are supported by eye movement data in ten languages and consider four models, spanning from 564M to 4.5B parameters.",
}
| Neural language models are increasingly valued in computational psycholinguistics, due to their ability to provide conditional probability distributions over the lexicon that are predictive of human processing times. Given the vast array of available models, it is of both theoretical and methodological importance to assess what features of a model influence its psychometric quality. In this work we focus on parameter size, showing that larger Transformer-based language models generate probabilistic estimates that are less predictive of early eye-tracking measurements reflecting lexical access and early semantic integration. However, relatively bigger models show an advantage in capturing late eye-tracking measurements that reflect the full semantic and syntactic integration of a word into the current language context. Our results are supported by eye movement data in ten languages and consider four models, spanning from 564M to 4.5B parameters. | [
"de Varda, Andrea",
"Marelli, Marco"
] | Scaling in Cognitive Modelling: a Multilingual Approach to Human Reading Times | acl-short.14 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.15.bib | https://aclanthology.org/2023.acl-short.15/ | @inproceedings{rai-etal-2023-improving,
title = "Improving Generalization in Language Model-based Text-to-{SQL} Semantic Parsing: Two Simple Semantic Boundary-based Techniques",
author = "Rai, Daking and
Wang, Bailin and
Zhou, Yilun and
Yao, Ziyu",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.15",
doi = "10.18653/v1/2023.acl-short.15",
pages = "150--160",
abstract = "Compositional and domain generalization present significant challenges in semantic parsing, even for state-of-the-art semantic parsers based on pre-trained language models (LMs). In this study, we empirically investigate improving an LM{'}s generalization in semantic parsing with two simple techniques: at the token level, we introduce a token preprocessing method to preserve the semantic boundaries of tokens produced by LM tokenizers; at the sequence level, we propose to use special tokens to mark the boundaries of components aligned between input and output. Our experimental results on two text-to-SQL semantic parsing datasets show that our token preprocessing, although simple, can substantially improve the LM performance on both types of generalization, and our component boundary marking method is particularly helpful for compositional generalization.",
}
| Compositional and domain generalization present significant challenges in semantic parsing, even for state-of-the-art semantic parsers based on pre-trained language models (LMs). In this study, we empirically investigate improving an LM{'}s generalization in semantic parsing with two simple techniques: at the token level, we introduce a token preprocessing method to preserve the semantic boundaries of tokens produced by LM tokenizers; at the sequence level, we propose to use special tokens to mark the boundaries of components aligned between input and output. Our experimental results on two text-to-SQL semantic parsing datasets show that our token preprocessing, although simple, can substantially improve the LM performance on both types of generalization, and our component boundary marking method is particularly helpful for compositional generalization. | [
"Rai, Daking",
"Wang, Bailin",
"Zhou, Yilun",
"Yao, Ziyu"
] | Improving Generalization in Language Model-based Text-to-SQL Semantic Parsing: Two Simple Semantic Boundary-based Techniques | acl-short.15 | Poster | 2305.17378 | [
"https://github.com/dakingrai/ood-generalization-semantic-boundary-techniques"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.16.bib | https://aclanthology.org/2023.acl-short.16/ | @inproceedings{li-etal-2023-hipool,
title = "{H}i{P}ool: Modeling Long Documents Using Graph Neural Networks",
author = "Li, Irene and
Feng, Aosong and
Radev, Dragomir and
Ying, Rex",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.16",
doi = "10.18653/v1/2023.acl-short.16",
pages = "161--171",
abstract = "Encoding long sequences in Natural Language Processing (NLP) is a challenging problem. Though recent pretraining language models achieve satisfying performances in many NLP tasks, they are still restricted by a pre-defined maximum length, making them challenging to be extended to longer sequences. So some recent works utilize hierarchies to model long sequences. However, most of them apply sequential models for upper hierarchies, suffering from long dependency issues. In this paper, we alleviate these issues through a graph-based method. We first chunk the sequence with a fixed length to model the sentence-level information. We then leverage graphs to model intra- and cross-sentence correlations with a new attention mechanism. Additionally, due to limited standard benchmarks for long document classification (LDC), we propose a new challenging benchmark, totaling six datasets with up to 53k samples and 4034 average tokens{'} length. Evaluation shows our model surpasses competitive baselines by 2.6{\%} in F1 score, and 4.8{\%} on the longest sequence dataset. Our method is shown to outperform hierarchical sequential models with better performance and scalability, especially for longer sequences.",
}
| Encoding long sequences in Natural Language Processing (NLP) is a challenging problem. Though recent pretraining language models achieve satisfying performances in many NLP tasks, they are still restricted by a pre-defined maximum length, making them challenging to be extended to longer sequences. So some recent works utilize hierarchies to model long sequences. However, most of them apply sequential models for upper hierarchies, suffering from long dependency issues. In this paper, we alleviate these issues through a graph-based method. We first chunk the sequence with a fixed length to model the sentence-level information. We then leverage graphs to model intra- and cross-sentence correlations with a new attention mechanism. Additionally, due to limited standard benchmarks for long document classification (LDC), we propose a new challenging benchmark, totaling six datasets with up to 53k samples and 4034 average tokens{'} length. Evaluation shows our model surpasses competitive baselines by 2.6{\%} in F1 score, and 4.8{\%} on the longest sequence dataset. Our method is shown to outperform hierarchical sequential models with better performance and scalability, especially for longer sequences. | [
"Li, Irene",
"Feng, Aosong",
"Radev, Dragomir",
"Ying, Rex"
] | HiPool: Modeling Long Documents Using Graph Neural Networks | acl-short.16 | Poster | 2305.03319 | [
"https://github.com/irenezihuili/hipool"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.17.bib | https://aclanthology.org/2023.acl-short.17/ | @inproceedings{yoder-etal-2023-weakly,
title = "A Weakly Supervised Classifier and Dataset of White Supremacist Language",
author = "Yoder, Michael and
Diab, Ahmad and
Brown, David and
Carley, Kathleen",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.17",
doi = "10.18653/v1/2023.acl-short.17",
pages = "172--185",
abstract = "We present a dataset and classifier for detecting the language of white supremacist extremism, a growing issue in online hate speech. Our weakly supervised classifier is trained on large datasets of text from explicitly white supremacist domains paired with neutral and anti-racist data from similar domains. We demonstrate that this approach improves generalization performance to new domains. Incorporating anti-racist texts as counterexamples to white supremacist language mitigates bias.",
}
| We present a dataset and classifier for detecting the language of white supremacist extremism, a growing issue in online hate speech. Our weakly supervised classifier is trained on large datasets of text from explicitly white supremacist domains paired with neutral and anti-racist data from similar domains. We demonstrate that this approach improves generalization performance to new domains. Incorporating anti-racist texts as counterexamples to white supremacist language mitigates bias. | [
"Yoder, Michael",
"Diab, Ahmad",
"Brown, David",
"Carley, Kathleen"
] | A Weakly Supervised Classifier and Dataset of White Supremacist Language | acl-short.17 | Poster | 2306.15732 | [
"https://github.com/michaelmilleryoder/white_supremacist_lang"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.18.bib | https://aclanthology.org/2023.acl-short.18/ | @inproceedings{liu-etal-2023-bolt,
title = "{BOLT}: Fast Energy-based Controlled Text Generation with Tunable Biases",
author = "Liu, Xin and
Khalifa, Muhammad and
Wang, Lu",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.18",
doi = "10.18653/v1/2023.acl-short.18",
pages = "186--200",
abstract = "Energy-based models (EBMs) have gained popularity for controlled text generation due to their high applicability to a wide range of constraints. However, sampling from EBMs is non-trivial, as it often requires a large number of iterations to converge to plausible text, which slows down the decoding process and makes it less practical for real-world applications. In this work, we propose BOLT, which relies on tunable biases to directly adjust the language model{'}s output logits. Unlike prior work, BOLT maintains the generator{'}s autoregressive nature to assert a strong control on token-wise conditional dependencies and overall fluency, and thus converges faster. When compared with state-of-the-arts on controlled generation tasks using both soft constraints (e.g., sentiment control) and hard constraints (e.g., keyword-guided topic control), BOLT demonstrates significantly improved efficiency and fluency. On sentiment control, BOLT is 7x faster than competitive baselines, and more fluent in 74.4{\%} of the evaluation samples according to human judges.",
}
| Energy-based models (EBMs) have gained popularity for controlled text generation due to their high applicability to a wide range of constraints. However, sampling from EBMs is non-trivial, as it often requires a large number of iterations to converge to plausible text, which slows down the decoding process and makes it less practical for real-world applications. In this work, we propose BOLT, which relies on tunable biases to directly adjust the language model{'}s output logits. Unlike prior work, BOLT maintains the generator{'}s autoregressive nature to assert a strong control on token-wise conditional dependencies and overall fluency, and thus converges faster. When compared with state-of-the-arts on controlled generation tasks using both soft constraints (e.g., sentiment control) and hard constraints (e.g., keyword-guided topic control), BOLT demonstrates significantly improved efficiency and fluency. On sentiment control, BOLT is 7x faster than competitive baselines, and more fluent in 74.4{\%} of the evaluation samples according to human judges. | [
"Liu, Xin",
"Khalifa, Muhammad",
"Wang, Lu"
] | BOLT: Fast Energy-based Controlled Text Generation with Tunable Biases | acl-short.18 | Poster | 2305.12018 | [
"https://github.com/launchnlp/bolt"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.19.bib | https://aclanthology.org/2023.acl-short.19/ | @inproceedings{mittal-etal-2023-mokb6,
title = "m{OKB}6: A Multilingual Open Knowledge Base Completion Benchmark",
author = "Mittal, Shubham and
Kolluru, Keshav and
Chakrabarti, Soumen and
{Mausam}",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.19",
doi = "10.18653/v1/2023.acl-short.19",
pages = "201--214",
abstract = "Automated completion of open knowledge bases (Open KBs), which are constructed from triples of the form (subject phrase, relation phrase, object phrase), obtained via open information extraction (Open IE) system, are useful for discovering novel facts that may not be directly present in the text. However, research in Open KB completion (Open KBC) has so far been limited to resource-rich languages like English. Using the latest advances in multilingual Open IE, we construct the first multilingual Open KBC dataset, called mOKB6, containing facts from Wikipedia in six languages (including English). Improvingthe previous Open KB construction pipeline by doing multilingual coreference resolution andkeeping only entity-linked triples, we create a dense Open KB. We experiment with several models for the task and observe a consistent benefit of combining languages with the help of shared embedding space as well as translations of facts. We also observe that current multilingual models struggle to remember facts seen in languages of different scripts.",
}
| Automated completion of open knowledge bases (Open KBs), which are constructed from triples of the form (subject phrase, relation phrase, object phrase), obtained via open information extraction (Open IE) system, are useful for discovering novel facts that may not be directly present in the text. However, research in Open KB completion (Open KBC) has so far been limited to resource-rich languages like English. Using the latest advances in multilingual Open IE, we construct the first multilingual Open KBC dataset, called mOKB6, containing facts from Wikipedia in six languages (including English). Improvingthe previous Open KB construction pipeline by doing multilingual coreference resolution andkeeping only entity-linked triples, we create a dense Open KB. We experiment with several models for the task and observe a consistent benefit of combining languages with the help of shared embedding space as well as translations of facts. We also observe that current multilingual models struggle to remember facts seen in languages of different scripts. | [
"Mittal, Shubham",
"Kolluru, Keshav",
"Chakrabarti, Soumen",
"{Mausam}"
] | mOKB6: A Multilingual Open Knowledge Base Completion Benchmark | acl-short.19 | Poster | 2211.06959 | [
"https://github.com/dair-iitd/mokb6"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.20.bib | https://aclanthology.org/2023.acl-short.20/ | @inproceedings{rabin-etal-2023-covering,
title = "Covering Uncommon Ground: Gap-Focused Question Generation for Answer Assessment",
author = "Rabin, Roni and
Djerbetian, Alexandre and
Engelberg, Roee and
Hackmon, Lidan and
Elidan, Gal and
Tsarfaty, Reut and
Globerson, Amir",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.20",
doi = "10.18653/v1/2023.acl-short.20",
pages = "215--227",
abstract = "Human communication often involves information gaps between the interlocutors. For example, in an educational dialogue a student often provides an answer that is incomplete, and there is a gap between this answer and the perfect one expected by the teacher. Successful dialogue then hinges on the teacher asking about this gap in an effective manner, thus creating a rich and interactive educational experience. We focus on the problem of generating such gap-focused questions (GFQs) automatically. We define the task, highlight key desired aspects of a good GFQ, and propose a model that satisfies these. Finally, we provide an evaluation by human annotators of our generated questions compared against human generated ones, demonstrating competitive performance.",
}
| Human communication often involves information gaps between the interlocutors. For example, in an educational dialogue a student often provides an answer that is incomplete, and there is a gap between this answer and the perfect one expected by the teacher. Successful dialogue then hinges on the teacher asking about this gap in an effective manner, thus creating a rich and interactive educational experience. We focus on the problem of generating such gap-focused questions (GFQs) automatically. We define the task, highlight key desired aspects of a good GFQ, and propose a model that satisfies these. Finally, we provide an evaluation by human annotators of our generated questions compared against human generated ones, demonstrating competitive performance. | [
"Rabin, Roni",
"Djerbetian, Alex",
"re",
"Engelberg, Roee",
"Hackmon, Lidan",
"Elidan, Gal",
"Tsarfaty, Reut",
"Globerson, Amir"
] | Covering Uncommon Ground: Gap-Focused Question Generation for Answer Assessment | acl-short.20 | Poster | 2307.03319 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.21.bib | https://aclanthology.org/2023.acl-short.21/ | @inproceedings{hallinan-etal-2023-detoxifying,
title = "Detoxifying Text with {M}a{RC}o: Controllable Revision with Experts and Anti-Experts",
author = "Hallinan, Skyler and
Liu, Alisa and
Choi, Yejin and
Sap, Maarten",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.21",
doi = "10.18653/v1/2023.acl-short.21",
pages = "228--242",
abstract = "Text detoxification has the potential to mitigate the harms of toxicity by rephrasing text to remove offensive meaning, but subtle toxicity remains challenging to tackle. We introduce MaRCo, a detoxification algorithm that combines controllable generation and text rewriting methods using a Product of Experts with autoencoder language models (LMs). MaRCo uses likelihoods under a non-toxic LM (expert) and a toxic LM (anti-expert) to find candidate words to mask and potentially replace. We evaluate our method on several subtle toxicity and microaggressions datasets, and show that it not only outperforms baselines on automatic metrics, but MaRCo{'}s rewrites are preferred 2.1 times more in human evaluation. Its applicability to instances of subtle toxicity is especially promising, demonstrating a path forward for addressing increasingly elusive online hate.",
}
| Text detoxification has the potential to mitigate the harms of toxicity by rephrasing text to remove offensive meaning, but subtle toxicity remains challenging to tackle. We introduce MaRCo, a detoxification algorithm that combines controllable generation and text rewriting methods using a Product of Experts with autoencoder language models (LMs). MaRCo uses likelihoods under a non-toxic LM (expert) and a toxic LM (anti-expert) to find candidate words to mask and potentially replace. We evaluate our method on several subtle toxicity and microaggressions datasets, and show that it not only outperforms baselines on automatic metrics, but MaRCo{'}s rewrites are preferred 2.1 times more in human evaluation. Its applicability to instances of subtle toxicity is especially promising, demonstrating a path forward for addressing increasingly elusive online hate. | [
"Hallinan, Skyler",
"Liu, Alisa",
"Choi, Yejin",
"Sap, Maarten"
] | Detoxifying Text with MaRCo: Controllable Revision with Experts and Anti-Experts | acl-short.21 | Poster | 2212.10543 | [
"https://github.com/shallinan1/marcodetoxification"
] | https://huggingface.co/papers/2212.10543 | 1 | 0 | 0 | 4 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-short.22.bib | https://aclanthology.org/2023.acl-short.22/ | @inproceedings{meister-etal-2023-natural,
title = "A Natural Bias for Language Generation Models",
author = "Meister, Clara and
Stokowiec, Wojciech and
Pimentel, Tiago and
Yu, Lei and
Rimell, Laura and
Kuncoro, Adhiguna",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.22",
doi = "10.18653/v1/2023.acl-short.22",
pages = "243--255",
abstract = "After just a few hundred training updates, a standard probabilistic model for language generation has likely not yet learnt many semantic or syntactic rules of natural language, making it difficult to estimate the probability distribution over next tokens. Yet around this point, these models have identified a simple, loss-minimising behaviour: to output the unigram distribution of the target training corpus. The use of such a heuristic raises the question: Can we initialise our models with this behaviour and save precious compute resources and model capacity? Here we show that we can effectively endow standard neural language generation models with a separate module that reflects unigram frequency statistics as prior knowledge, simply by initialising the bias term in a model{'}s final linear layer with the log-unigram distribution. We use neural machine translation as a test bed for this simple technique and observe that it: (i) improves learning efficiency; (ii) achieves better overall performance; and perhaps most importantly (iii) appears to disentangle strong frequency effects by encouraging the model to specialise in non-frequency-related aspects of language.",
}
| After just a few hundred training updates, a standard probabilistic model for language generation has likely not yet learnt many semantic or syntactic rules of natural language, making it difficult to estimate the probability distribution over next tokens. Yet around this point, these models have identified a simple, loss-minimising behaviour: to output the unigram distribution of the target training corpus. The use of such a heuristic raises the question: Can we initialise our models with this behaviour and save precious compute resources and model capacity? Here we show that we can effectively endow standard neural language generation models with a separate module that reflects unigram frequency statistics as prior knowledge, simply by initialising the bias term in a model{'}s final linear layer with the log-unigram distribution. We use neural machine translation as a test bed for this simple technique and observe that it: (i) improves learning efficiency; (ii) achieves better overall performance; and perhaps most importantly (iii) appears to disentangle strong frequency effects by encouraging the model to specialise in non-frequency-related aspects of language. | [
"Meister, Clara",
"Stokowiec, Wojciech",
"Pimentel, Tiago",
"Yu, Lei",
"Rimell, Laura",
"Kuncoro, Adhiguna"
] | A Natural Bias for Language Generation Models | acl-short.22 | Poster | 2212.09686 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.23.bib | https://aclanthology.org/2023.acl-short.23/ | @inproceedings{nandi-etal-2023-simple,
title = "Simple Augmentations of Logical Rules for Neuro-Symbolic Knowledge Graph Completion",
author = "Nandi, Ananjan and
Kaur, Navdeep and
Singla, Parag and
{Mausam}",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.23",
doi = "10.18653/v1/2023.acl-short.23",
pages = "256--269",
abstract = "High-quality and high-coverage rule sets are imperative to the success of Neuro-Symbolic Knowledge Graph Completion (NS-KGC) models, because they form the basis of all symbolic inferences. Recent literature builds neural models for generating rule sets, however, preliminary experiments show that they struggle with maintaining high coverage. In this work, we suggest three simple augmentations to existing rule sets: (1) transforming rules to their abductive forms, (2) generating equivalent rules that use inverse forms of constituent relations and (3) random walks that propose new rules. Finally, we prune potentially low quality rules. Experiments over four datasets and five ruleset-baseline settings suggest that these simple augmentations consistently improve results, and obtain up to 7.1 pt MRR and 8.5 pt Hits@1 gains over using rules without augmentations.",
}
| High-quality and high-coverage rule sets are imperative to the success of Neuro-Symbolic Knowledge Graph Completion (NS-KGC) models, because they form the basis of all symbolic inferences. Recent literature builds neural models for generating rule sets, however, preliminary experiments show that they struggle with maintaining high coverage. In this work, we suggest three simple augmentations to existing rule sets: (1) transforming rules to their abductive forms, (2) generating equivalent rules that use inverse forms of constituent relations and (3) random walks that propose new rules. Finally, we prune potentially low quality rules. Experiments over four datasets and five ruleset-baseline settings suggest that these simple augmentations consistently improve results, and obtain up to 7.1 pt MRR and 8.5 pt Hits@1 gains over using rules without augmentations. | [
"N",
"i, Ananjan",
"Kaur, Navdeep",
"Singla, Parag",
"{Mausam}"
] | Simple Augmentations of Logical Rules for Neuro-Symbolic Knowledge Graph Completion | acl-short.23 | Poster | 2407.01994 | [
"https://github.com/dair-iitd/NS-KGC-AUG"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.24.bib | https://aclanthology.org/2023.acl-short.24/ | @inproceedings{lv-etal-2023-parameter,
title = "Parameter-efficient Weight Ensembling Facilitates Task-level Knowledge Transfer",
author = "Lv, Xingtai and
Ding, Ning and
Qin, Yujia and
Liu, Zhiyuan and
Sun, Maosong",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.24",
doi = "10.18653/v1/2023.acl-short.24",
pages = "270--282",
abstract = "Recent studies show that large-scale pre-trained language models could be efficaciously adapted to particular tasks in a parameter-efficient manner. The trained lightweight set of parameters, such as adapters, can be easily stored and shared as a capability equipped with the corresponding models. Owning many lightweight parameters, we focus on transferring them between tasks to acquire an improvement in performance of new tasks, the key point of which is to obtain the similarity between tasks. In this paper, we explore 5 parameter-efficient weight ensembling methods to achieve such transferability and verify the effectiveness of them. These methods extract the information of datasets and trained lightweight parameters from different perspectives to obtain the similarity between tasks, and weight the existing lightweight parameters according to the comparability to acquire a suitable module for the initialization of new tasks. We apply them to three parameter-efficient tuning methods and test them on a wide set of downstream tasks. Experimental results show that our methods show an improvement of 5{\%}{\textasciitilde}8{\%} over baselines and could largely facilitate task-level knowledge transfer.",
}
| Recent studies show that large-scale pre-trained language models could be efficaciously adapted to particular tasks in a parameter-efficient manner. The trained lightweight set of parameters, such as adapters, can be easily stored and shared as a capability equipped with the corresponding models. Owning many lightweight parameters, we focus on transferring them between tasks to acquire an improvement in performance of new tasks, the key point of which is to obtain the similarity between tasks. In this paper, we explore 5 parameter-efficient weight ensembling methods to achieve such transferability and verify the effectiveness of them. These methods extract the information of datasets and trained lightweight parameters from different perspectives to obtain the similarity between tasks, and weight the existing lightweight parameters according to the comparability to acquire a suitable module for the initialization of new tasks. We apply them to three parameter-efficient tuning methods and test them on a wide set of downstream tasks. Experimental results show that our methods show an improvement of 5{\%}{\textasciitilde}8{\%} over baselines and could largely facilitate task-level knowledge transfer. | [
"Lv, Xingtai",
"Ding, Ning",
"Qin, Yujia",
"Liu, Zhiyuan",
"Sun, Maosong"
] | Parameter-efficient Weight Ensembling Facilitates Task-level Knowledge Transfer | acl-short.24 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.25.bib | https://aclanthology.org/2023.acl-short.25/ | @inproceedings{atanasova-etal-2023-faithfulness,
title = "Faithfulness Tests for Natural Language Explanations",
author = "Atanasova, Pepa and
Camburu, Oana-Maria and
Lioma, Christina and
Lukasiewicz, Thomas and
Simonsen, Jakob Grue and
Augenstein, Isabelle",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.25",
doi = "10.18653/v1/2023.acl-short.25",
pages = "283--294",
abstract = "Explanations of neural models aim to reveal a model{'}s decision-making process for its predictions. However, recent work shows that current methods giving explanations such as saliency maps or counterfactuals can be misleading, as they are prone to present reasons that are unfaithful to the model{'}s inner workings. This work explores the challenging question of evaluating the faithfulness of natural language explanations (NLEs). To this end, we present two tests. First, we propose a counterfactual input editor for inserting reasons that lead to counterfactual predictions but are not reflected by the NLEs. Second, we reconstruct inputs from the reasons stated in the generated NLEs and check how often they lead to the same predictions. Our tests can evaluate emerging NLE models, proving a fundamental tool in the development of faithful NLEs.",
}
| Explanations of neural models aim to reveal a model{'}s decision-making process for its predictions. However, recent work shows that current methods giving explanations such as saliency maps or counterfactuals can be misleading, as they are prone to present reasons that are unfaithful to the model{'}s inner workings. This work explores the challenging question of evaluating the faithfulness of natural language explanations (NLEs). To this end, we present two tests. First, we propose a counterfactual input editor for inserting reasons that lead to counterfactual predictions but are not reflected by the NLEs. Second, we reconstruct inputs from the reasons stated in the generated NLEs and check how often they lead to the same predictions. Our tests can evaluate emerging NLE models, proving a fundamental tool in the development of faithful NLEs. | [
"Atanasova, Pepa",
"Camburu, Oana-Maria",
"Lioma, Christina",
"Lukasiewicz, Thomas",
"Simonsen, Jakob Grue",
"Augenstein, Isabelle"
] | Faithfulness Tests for Natural Language Explanations | acl-short.25 | Oral | 2305.18029 | [
"https://github.com/copenlu/nle_faithfulness"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.26.bib | https://aclanthology.org/2023.acl-short.26/ | @inproceedings{zandie-etal-2023-cogen,
title = "{COGEN}: Abductive Commonsense Language Generation",
author = "Zandie, Rohola and
Shekhar, Diwanshu and
Mahoor, Mohammad",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.26",
doi = "10.18653/v1/2023.acl-short.26",
pages = "295--302",
abstract = "Reasoning is one of the most important elements in achieving Artificial General Intelligence (AGI), specifically when it comes to Abductive and counterfactual reasoning. In order to introduce these capabilities of reasoning in Natural Language Processing (NLP) models, there have been recent advances towards training NLP models to better perform on two main tasks - Abductive Natural Language Inference (alphaNLI) and Abductive Natural Language Generation Task (alphaNLG). This paper proposes CoGen, a model for both alphaNLI and alphaNLG tasks that employ a novel approach of combining the temporal commonsense reasoning for each observation (before and after a real hypothesis) from pre-trained models with contextual filtering for training. Additionally, we use state-of-the-art semantic entailment to filter out the contradictory hypothesis during the inference. Our experimental results show that CoGen outperforms current models and set a new state of the art in regards to alphaNLI and alphaNLG tasks. We make the source code of CoGen model publicly available for reproducibility and to facilitate relevant future research.",
}
| Reasoning is one of the most important elements in achieving Artificial General Intelligence (AGI), specifically when it comes to Abductive and counterfactual reasoning. In order to introduce these capabilities of reasoning in Natural Language Processing (NLP) models, there have been recent advances towards training NLP models to better perform on two main tasks - Abductive Natural Language Inference (alphaNLI) and Abductive Natural Language Generation Task (alphaNLG). This paper proposes CoGen, a model for both alphaNLI and alphaNLG tasks that employ a novel approach of combining the temporal commonsense reasoning for each observation (before and after a real hypothesis) from pre-trained models with contextual filtering for training. Additionally, we use state-of-the-art semantic entailment to filter out the contradictory hypothesis during the inference. Our experimental results show that CoGen outperforms current models and set a new state of the art in regards to alphaNLI and alphaNLG tasks. We make the source code of CoGen model publicly available for reproducibility and to facilitate relevant future research. | [
"Z",
"ie, Rohola",
"Shekhar, Diwanshu",
"Mahoor, Mohammad"
] | COGEN: Abductive Commonsense Language Generation | acl-short.26 | Oral | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.27.bib | https://aclanthology.org/2023.acl-short.27/ | @inproceedings{hu-etal-2023-multimodal,
title = "Multimodal Relation Extraction with Cross-Modal Retrieval and Synthesis",
author = "Hu, Xuming and
Guo, Zhijiang and
Teng, Zhiyang and
King, Irwin and
Yu, Philip S.",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.27",
doi = "10.18653/v1/2023.acl-short.27",
pages = "303--311",
abstract = "Multimodal relation extraction (MRE) is the task of identifying the semantic relationships between two entities based on the context of the sentence image pair. Existing retrieval-augmented approaches mainly focused on modeling the retrieved textual knowledge, but this may not be able to accurately identify complex relations. To improve the prediction, this research proposes to retrieve textual and visual evidence based on the object, sentence, and whole image. We further develop a novel approach to synthesize the object-level, image-level, and sentence-level information for better reasoning between the same and different modalities. Extensive experiments and analyses show that the proposed method is able to effectively select and compare evidence across modalities and significantly outperforms state-of-the-art models.",
}
| Multimodal relation extraction (MRE) is the task of identifying the semantic relationships between two entities based on the context of the sentence image pair. Existing retrieval-augmented approaches mainly focused on modeling the retrieved textual knowledge, but this may not be able to accurately identify complex relations. To improve the prediction, this research proposes to retrieve textual and visual evidence based on the object, sentence, and whole image. We further develop a novel approach to synthesize the object-level, image-level, and sentence-level information for better reasoning between the same and different modalities. Extensive experiments and analyses show that the proposed method is able to effectively select and compare evidence across modalities and significantly outperforms state-of-the-art models. | [
"Hu, Xuming",
"Guo, Zhijiang",
"Teng, Zhiyang",
"King, Irwin",
"Yu, Philip S."
] | Multimodal Relation Extraction with Cross-Modal Retrieval and Synthesis | acl-short.27 | Poster | 2305.16166 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.28.bib | https://aclanthology.org/2023.acl-short.28/ | @inproceedings{harrigian-etal-2023-characterization,
title = "Characterization of Stigmatizing Language in Medical Records",
author = "Harrigian, Keith and
Zirikly, Ayah and
Chee, Brant and
Ahmad, Alya and
Links, Anne and
Saha, Somnath and
Beach, Mary Catherine and
Dredze, Mark",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.28",
doi = "10.18653/v1/2023.acl-short.28",
pages = "312--329",
abstract = "Widespread disparities in clinical outcomes exist between different demographic groups in the United States. A new line of work in medical sociology has demonstrated physicians often use stigmatizing language in electronic medical records within certain groups, such as black patients, which may exacerbate disparities. In this study, we characterize these instances at scale using a series of domain-informed NLP techniques. We highlight important differences between this task and analogous bias-related tasks studied within the NLP community (e.g., classifying microaggressions). Our study establishes a foundation for NLP researchers to contribute timely insights to a problem domain brought to the forefront by recent legislation regarding clinical documentation transparency. We release data, code, and models.",
}
| Widespread disparities in clinical outcomes exist between different demographic groups in the United States. A new line of work in medical sociology has demonstrated physicians often use stigmatizing language in electronic medical records within certain groups, such as black patients, which may exacerbate disparities. In this study, we characterize these instances at scale using a series of domain-informed NLP techniques. We highlight important differences between this task and analogous bias-related tasks studied within the NLP community (e.g., classifying microaggressions). Our study establishes a foundation for NLP researchers to contribute timely insights to a problem domain brought to the forefront by recent legislation regarding clinical documentation transparency. We release data, code, and models. | [
"Harrigian, Keith",
"Zirikly, Ayah",
"Chee, Brant",
"Ahmad, Alya",
"Links, Anne",
"Saha, Somnath",
"Beach, Mary Catherine",
"Dredze, Mark"
] | Characterization of Stigmatizing Language in Medical Records | acl-short.28 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.29.bib | https://aclanthology.org/2023.acl-short.29/ | @inproceedings{varab-xu-2023-abstractive,
title = "Abstractive Summarizers are Excellent Extractive Summarizers",
author = "Varab, Daniel and
Xu, Yumo",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.29",
doi = "10.18653/v1/2023.acl-short.29",
pages = "330--339",
abstract = "Extractive and abstractive summarization designs have historically been fragmented, limiting the benefits that often arise from compatible model architectures. In this paper, we explore the potential synergies of modeling extractive summarization with an abstractive summarization system and propose three novel inference algorithms using the sequence-to-sequence architecture. We evaluate them on the CNN {\&} Dailymail dataset and show that recent advancements in abstractive system designs enable abstractive systems to not only compete, but even surpass the performance of extractive systems with custom architectures. To our surprise, abstractive systems achieve this without being exposed to extractive oracle summaries and, therefore, for the first time allow a single model to produce both abstractive and extractive summaries. This evidence questions our fundamental understanding of extractive system design, and the necessity for extractive labels while pathing the way for promising research directions in hybrid models.",
}
| Extractive and abstractive summarization designs have historically been fragmented, limiting the benefits that often arise from compatible model architectures. In this paper, we explore the potential synergies of modeling extractive summarization with an abstractive summarization system and propose three novel inference algorithms using the sequence-to-sequence architecture. We evaluate them on the CNN {\&} Dailymail dataset and show that recent advancements in abstractive system designs enable abstractive systems to not only compete, but even surpass the performance of extractive systems with custom architectures. To our surprise, abstractive systems achieve this without being exposed to extractive oracle summaries and, therefore, for the first time allow a single model to produce both abstractive and extractive summaries. This evidence questions our fundamental understanding of extractive system design, and the necessity for extractive labels while pathing the way for promising research directions in hybrid models. | [
"Varab, Daniel",
"Xu, Yumo"
] | Abstractive Summarizers are Excellent Extractive Summarizers | acl-short.29 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.30.bib | https://aclanthology.org/2023.acl-short.30/ | @inproceedings{thakur-etal-2023-language,
title = "Language Models Get a Gender Makeover: Mitigating Gender Bias with Few-Shot Data Interventions",
author = "Thakur, Himanshu and
Jain, Atishay and
Vaddamanu, Praneetha and
Liang, Paul Pu and
Morency, Louis-Philippe",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.30",
doi = "10.18653/v1/2023.acl-short.30",
pages = "340--351",
abstract = "Societal biases present in pre-trained large language models are a critical issue as these models have been shown to propagate biases in countless downstream applications, rendering them unfair towards specific groups of people. Since large-scale retraining of these models from scratch is both time and compute-expensive, a variety of approaches have been previously proposed that de-bias a pre-trained model. While the majority of current state-of-the-art debiasing methods focus on changes to the training regime, in this paper, we propose data intervention strategies as a powerful yet simple technique to reduce gender bias in pre-trained models. Specifically, we empirically show that by fine-tuning a pre-trained model on only 10 debiased (intervened) training examples, the tendency to favor any gender is significantly reduced. Since our proposed method only needs a few training examples, we argue that our few-shot de-biasing approach is highly feasible and practical. Through extensive experimentation, we show that our de-biasing technique performs better than competitive state-of-the-art baselines with minimal loss in language modeling ability.",
}
| Societal biases present in pre-trained large language models are a critical issue as these models have been shown to propagate biases in countless downstream applications, rendering them unfair towards specific groups of people. Since large-scale retraining of these models from scratch is both time and compute-expensive, a variety of approaches have been previously proposed that de-bias a pre-trained model. While the majority of current state-of-the-art debiasing methods focus on changes to the training regime, in this paper, we propose data intervention strategies as a powerful yet simple technique to reduce gender bias in pre-trained models. Specifically, we empirically show that by fine-tuning a pre-trained model on only 10 debiased (intervened) training examples, the tendency to favor any gender is significantly reduced. Since our proposed method only needs a few training examples, we argue that our few-shot de-biasing approach is highly feasible and practical. Through extensive experimentation, we show that our de-biasing technique performs better than competitive state-of-the-art baselines with minimal loss in language modeling ability. | [
"Thakur, Himanshu",
"Jain, Atishay",
"Vaddamanu, Praneetha",
"Liang, Paul Pu",
"Morency, Louis-Philippe"
] | Language Models Get a Gender Makeover: Mitigating Gender Bias with Few-Shot Data Interventions | acl-short.30 | Poster | 2306.04597 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.31.bib | https://aclanthology.org/2023.acl-short.31/ | @inproceedings{chi-etal-2023-plue,
title = "{PLUE}: Language Understanding Evaluation Benchmark for Privacy Policies in {E}nglish",
author = "Chi, Jianfeng and
Ahmad, Wasi Uddin and
Tian, Yuan and
Chang, Kai-Wei",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.31",
doi = "10.18653/v1/2023.acl-short.31",
pages = "352--365",
abstract = "Privacy policies provide individuals with information about their rights and how their personal information is handled. Natural language understanding (NLU) technologies can support individuals and practitioners to understand better privacy practices described in lengthy and complex documents. However, existing efforts that use NLU technologies are limited by processing the language in a way exclusive to a single task focusing on certain privacy practices. To this end, we introduce the Privacy Policy Language Understanding Evaluation (PLUE) benchmark, a multi-task benchmark for evaluating the privacy policy language understanding across various tasks. We also collect a large corpus of privacy policies to enable privacy policy domain-specific language model pre-training. We evaluate several generic pre-trained language models and continue pre-training them on the collected corpus. We demonstrate that domain-specific continual pre-training offers performance improvements across all tasks. The code and models are released at \url{https://github.com/JFChi/PLUE}.",
}
| Privacy policies provide individuals with information about their rights and how their personal information is handled. Natural language understanding (NLU) technologies can support individuals and practitioners to understand better privacy practices described in lengthy and complex documents. However, existing efforts that use NLU technologies are limited by processing the language in a way exclusive to a single task focusing on certain privacy practices. To this end, we introduce the Privacy Policy Language Understanding Evaluation (PLUE) benchmark, a multi-task benchmark for evaluating the privacy policy language understanding across various tasks. We also collect a large corpus of privacy policies to enable privacy policy domain-specific language model pre-training. We evaluate several generic pre-trained language models and continue pre-training them on the collected corpus. We demonstrate that domain-specific continual pre-training offers performance improvements across all tasks. The code and models are released at \url{https://github.com/JFChi/PLUE}. | [
"Chi, Jianfeng",
"Ahmad, Wasi Uddin",
"Tian, Yuan",
"Chang, Kai-Wei"
] | PLUE: Language Understanding Evaluation Benchmark for Privacy Policies in English | acl-short.31 | Poster | 2212.10011 | [
"https://github.com/jfchi/plue"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.32.bib | https://aclanthology.org/2023.acl-short.32/ | @inproceedings{karoui-etal-2023-stop,
title = "Stop Pre-Training: Adapt Visual-Language Models to Unseen Languages",
author = "Karoui, Yasmine and
Lebret, R{\'e}mi and
Foroutan Eghlidi, Negar and
Aberer, Karl",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.32",
doi = "10.18653/v1/2023.acl-short.32",
pages = "366--375",
abstract = "Vision-Language Pre-training (VLP) has advanced the performance of many vision-language tasks, such as image-text retrieval, visual entailment, and visual reasoning. The pre-training mostly utilizes lexical databases and image queries in English. Previous work has demonstrated that the pre-training in English does not transfer well to other languages in a zero-shot setting. However, multilingual pre-trained language models (MPLM) have excelled at a variety of single-modal language tasks. In this paper, we propose a simple yet efficient approach to adapt VLP to unseen languages using MPLM.We utilize a cross-lingual contextualised token embeddings alignment approach to train text encoders for non-English languages. Our approach does not require image input and primarily uses machine translation, eliminating the need for target language data. Our evaluation across three distinct tasks (image-text retrieval, visual entailment, and natural language visual reasoning) demonstrates that this approach outperforms the state-of-the-art multilingual vision-language models without requiring large parallel corpora. Our code is available at \url{https://github.com/Yasminekaroui/CliCoTea}.",
}
| Vision-Language Pre-training (VLP) has advanced the performance of many vision-language tasks, such as image-text retrieval, visual entailment, and visual reasoning. The pre-training mostly utilizes lexical databases and image queries in English. Previous work has demonstrated that the pre-training in English does not transfer well to other languages in a zero-shot setting. However, multilingual pre-trained language models (MPLM) have excelled at a variety of single-modal language tasks. In this paper, we propose a simple yet efficient approach to adapt VLP to unseen languages using MPLM.We utilize a cross-lingual contextualised token embeddings alignment approach to train text encoders for non-English languages. Our approach does not require image input and primarily uses machine translation, eliminating the need for target language data. Our evaluation across three distinct tasks (image-text retrieval, visual entailment, and natural language visual reasoning) demonstrates that this approach outperforms the state-of-the-art multilingual vision-language models without requiring large parallel corpora. Our code is available at \url{https://github.com/Yasminekaroui/CliCoTea}. | [
"Karoui, Yasmine",
"Lebret, R{\\'e}mi",
"Foroutan Eghlidi, Negar",
"Aberer, Karl"
] | Stop Pre-Training: Adapt Visual-Language Models to Unseen Languages | acl-short.32 | Poster | 2306.16774 | [
"https://github.com/yasminekaroui/clicotea"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.33.bib | https://aclanthology.org/2023.acl-short.33/ | @inproceedings{he-etal-2023-buca,
title = "{BUCA}: A Binary Classification Approach to Unsupervised Commonsense Question Answering",
author = "He, Jie and
U, Simon and
Gutierrez-Basulto, Victor and
Pan, Jeff",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.33",
doi = "10.18653/v1/2023.acl-short.33",
pages = "376--387",
abstract = "Unsupervised commonsense reasoning (UCR) is becoming increasingly popular as the construction of commonsense reasoning datasets is expensive, and they are inevitably limited in their scope. A popular approach to UCR is to fine-tune language models with external knowledge (e.g., knowledge graphs), but this usually requires a large number of training examples. In this paper, we propose to transform the downstream multiple choice question answering task into a simpler binary classification task by ranking all candidate answers according to their reasonableness. To this end, for training the model, we convert the knowledge graph triples into reasonable and unreasonable texts. Extensive experimental results show the effectiveness of our approach on various multiple choice question answering benchmarks. Furthermore, compared with existing UCR approaches using KGs, ours is less data hungry.",
}
| Unsupervised commonsense reasoning (UCR) is becoming increasingly popular as the construction of commonsense reasoning datasets is expensive, and they are inevitably limited in their scope. A popular approach to UCR is to fine-tune language models with external knowledge (e.g., knowledge graphs), but this usually requires a large number of training examples. In this paper, we propose to transform the downstream multiple choice question answering task into a simpler binary classification task by ranking all candidate answers according to their reasonableness. To this end, for training the model, we convert the knowledge graph triples into reasonable and unreasonable texts. Extensive experimental results show the effectiveness of our approach on various multiple choice question answering benchmarks. Furthermore, compared with existing UCR approaches using KGs, ours is less data hungry. | [
"He, Jie",
"U, Simon",
"Gutierrez-Basulto, Victor",
"Pan, Jeff"
] | BUCA: A Binary Classification Approach to Unsupervised Commonsense Question Answering | acl-short.33 | Poster | 2305.15932 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.34.bib | https://aclanthology.org/2023.acl-short.34/ | @inproceedings{an-rudinger-2023-nichelle,
title = "Nichelle and Nancy: The Influence of Demographic Attributes and Tokenization Length on First Name Biases",
author = "An, Haozhe and
Rudinger, Rachel",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.34",
doi = "10.18653/v1/2023.acl-short.34",
pages = "388--401",
abstract = "Through the use of first name substitution experiments, prior research has demonstrated the tendency of social commonsense reasoning models to systematically exhibit social biases along the dimensions of race, ethnicity, and gender (An et al., 2023). Demographic attributes of first names, however, are strongly correlated with corpus frequency and tokenization length, which may influence model behavior independent of or in addition to demographic factors. In this paper, we conduct a new series of first name substitution experiments that measures the influence of these factors while controlling for the others. We find that demographic attributes of a name (race, ethnicity, and gender) and name tokenization length are both factors that systematically affect the behavior of social commonsense reasoning models.",
}
| Through the use of first name substitution experiments, prior research has demonstrated the tendency of social commonsense reasoning models to systematically exhibit social biases along the dimensions of race, ethnicity, and gender (An et al., 2023). Demographic attributes of first names, however, are strongly correlated with corpus frequency and tokenization length, which may influence model behavior independent of or in addition to demographic factors. In this paper, we conduct a new series of first name substitution experiments that measures the influence of these factors while controlling for the others. We find that demographic attributes of a name (race, ethnicity, and gender) and name tokenization length are both factors that systematically affect the behavior of social commonsense reasoning models. | [
"An, Haozhe",
"Rudinger, Rachel"
] | Nichelle and Nancy: The Influence of Demographic Attributes and Tokenization Length on First Name Biases | acl-short.34 | Poster | 2305.16577 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.35.bib | https://aclanthology.org/2023.acl-short.35/ | @inproceedings{ma-etal-2023-improving,
title = "Improving Syntactic Probing Correctness and Robustness with Control Tasks",
author = "Ma, Weicheng and
Wang, Brian and
Zhang, Hefan and
Wang, Lili and
Coto-Solano, Rolando and
Hassanpour, Saeed and
Vosoughi, Soroush",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.35",
doi = "10.18653/v1/2023.acl-short.35",
pages = "402--415",
abstract = "Syntactic probing methods have been used to examine whether and how pre-trained language models (PLMs) encode syntactic features. However, the probing methods are usually biased by the PLMs{'} memorization of common word co-occurrences, even if they do not form syntactic relations. This paper presents a random-word-substitution and random-label-matching control task to reduce these biases and improve the robustness of syntactic probing methods. Our control tasks are also shown to notably improve the consistency of probing results between different probing methods and make the methods more robust with respect to the text attributes of the probing instances. Our control tasks make syntactic probing methods better at reconstructing syntactic features and more generalizable to unseen text domains. Our experiments show that our proposed control tasks are effective on different PLMs, probing methods, and syntactic features.",
}
| Syntactic probing methods have been used to examine whether and how pre-trained language models (PLMs) encode syntactic features. However, the probing methods are usually biased by the PLMs{'} memorization of common word co-occurrences, even if they do not form syntactic relations. This paper presents a random-word-substitution and random-label-matching control task to reduce these biases and improve the robustness of syntactic probing methods. Our control tasks are also shown to notably improve the consistency of probing results between different probing methods and make the methods more robust with respect to the text attributes of the probing instances. Our control tasks make syntactic probing methods better at reconstructing syntactic features and more generalizable to unseen text domains. Our experiments show that our proposed control tasks are effective on different PLMs, probing methods, and syntactic features. | [
"Ma, Weicheng",
"Wang, Brian",
"Zhang, Hefan",
"Wang, Lili",
"Coto-Solano, Rol",
"o",
"Hassanpour, Saeed",
"Vosoughi, Soroush"
] | Improving Syntactic Probing Correctness and Robustness with Control Tasks | acl-short.35 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.36.bib | https://aclanthology.org/2023.acl-short.36/ | @inproceedings{arora-park-2023-split,
title = "Split-{NER}: Named Entity Recognition via Two Question-Answering-based Classifications",
author = "Arora, Jatin and
Park, Youngja",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.36",
doi = "10.18653/v1/2023.acl-short.36",
pages = "416--426",
abstract = "In this work, we address the NER problem by splitting it into two logical sub-tasks: (1) Span Detection which simply extracts entity mention spans irrespective of entity type; (2) Span Classification which classifies the spans into their entity types. Further, we formulate both sub-tasks as question-answering (QA) problems and produce two leaner models which can be optimized separately for each sub-task. Experiments with four cross-domain datasets demonstrate that this two-step approach is both effective and time efficient. Our system, SplitNER outperforms baselines on OntoNotes5.0, WNUT17 and a cybersecurity dataset and gives on-par performance on BioNLP13CG. In all cases, it achieves a significant reduction in training time compared to its QA baseline counterpart. The effectiveness of our system stems from fine-tuning the BERT model twice, separately for span detection and classification. The source code can be found at \url{https://github.com/c3sr/split-ner}.",
}
| In this work, we address the NER problem by splitting it into two logical sub-tasks: (1) Span Detection which simply extracts entity mention spans irrespective of entity type; (2) Span Classification which classifies the spans into their entity types. Further, we formulate both sub-tasks as question-answering (QA) problems and produce two leaner models which can be optimized separately for each sub-task. Experiments with four cross-domain datasets demonstrate that this two-step approach is both effective and time efficient. Our system, SplitNER outperforms baselines on OntoNotes5.0, WNUT17 and a cybersecurity dataset and gives on-par performance on BioNLP13CG. In all cases, it achieves a significant reduction in training time compared to its QA baseline counterpart. The effectiveness of our system stems from fine-tuning the BERT model twice, separately for span detection and classification. The source code can be found at \url{https://github.com/c3sr/split-ner}. | [
"Arora, Jatin",
"Park, Youngja"
] | Split-NER: Named Entity Recognition via Two Question-Answering-based Classifications | acl-short.36 | Poster | 2310.19942 | [
"https://github.com/c3sr/split-ner"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.37.bib | https://aclanthology.org/2023.acl-short.37/ | @inproceedings{peskoff-stewart-2023-credible,
title = "Credible without Credit: Domain Experts Assess Generative Language Models",
author = "Peskoff, Denis and
Stewart, Brandon",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.37",
doi = "10.18653/v1/2023.acl-short.37",
pages = "427--438",
abstract = "Language models have recently broken into the public consciousness with the release of the wildly popular ChatGPT. Commentators have argued that language models could replace search engines, make college essays obsolete, or even write academic research papers. All of these tasks rely on accuracy of specialized information which can be difficult to assess for non-experts. Using 10 domain experts across science and culture, we provide an initial assessment of the coherence, conciseness, accuracy, and sourcing of two language models across 100 expert-written questions. While we find the results are consistently cohesive and concise, we find that they are mixed in their accuracy. These results raise questions of the role language models should play in general-purpose and expert knowledge seeking.",
}
| Language models have recently broken into the public consciousness with the release of the wildly popular ChatGPT. Commentators have argued that language models could replace search engines, make college essays obsolete, or even write academic research papers. All of these tasks rely on accuracy of specialized information which can be difficult to assess for non-experts. Using 10 domain experts across science and culture, we provide an initial assessment of the coherence, conciseness, accuracy, and sourcing of two language models across 100 expert-written questions. While we find the results are consistently cohesive and concise, we find that they are mixed in their accuracy. These results raise questions of the role language models should play in general-purpose and expert knowledge seeking. | [
"Peskoff, Denis",
"Stewart, Br",
"on"
] | Credible without Credit: Domain Experts Assess Generative Language Models | acl-short.37 | Oral | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.38.bib | https://aclanthology.org/2023.acl-short.38/ | @inproceedings{murty-etal-2023-grokking,
title = "Grokking of Hierarchical Structure in Vanilla Transformers",
author = "Murty, Shikhar and
Sharma, Pratyusha and
Andreas, Jacob and
Manning, Christopher",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.38",
doi = "10.18653/v1/2023.acl-short.38",
pages = "439--448",
abstract = "For humans, language production and comprehension is sensitive to the hierarchical structure of sentences. In natural language processing, past work has questioned how effectively neural sequence models like transformers capture this hierarchical structure when generalizing to structurally novel inputs. We show that transformer language models can learn to generalize hierarchically after training for extremely long periods{---}far beyond the point when in-domain accuracy has saturated. We call this phenomenon structural grokking. On multiple datasets, structural grokking exhibits inverted U-shaped scaling in model depth: intermediate-depth models generalize better than both very deep and very shallow transformers. When analyzing the relationship between model-internal properties and grokking, we find that optimal depth for grokking can be identified using the tree-structuredness metric of CITATION. Overall, our work provides strong evidence that, with extended training, vanilla transformers discover and use hierarchical structure.",
}
| For humans, language production and comprehension is sensitive to the hierarchical structure of sentences. In natural language processing, past work has questioned how effectively neural sequence models like transformers capture this hierarchical structure when generalizing to structurally novel inputs. We show that transformer language models can learn to generalize hierarchically after training for extremely long periods{---}far beyond the point when in-domain accuracy has saturated. We call this phenomenon structural grokking. On multiple datasets, structural grokking exhibits inverted U-shaped scaling in model depth: intermediate-depth models generalize better than both very deep and very shallow transformers. When analyzing the relationship between model-internal properties and grokking, we find that optimal depth for grokking can be identified using the tree-structuredness metric of CITATION. Overall, our work provides strong evidence that, with extended training, vanilla transformers discover and use hierarchical structure. | [
"Murty, Shikhar",
"Sharma, Pratyusha",
"Andreas, Jacob",
"Manning, Christopher"
] | Grokking of Hierarchical Structure in Vanilla Transformers | acl-short.38 | Poster | 2305.18741 | [
"https://github.com/murtyshikhar/structural-grokking"
] | https://huggingface.co/papers/2305.18741 | 1 | 0 | 0 | 4 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-short.39.bib | https://aclanthology.org/2023.acl-short.39/ | @inproceedings{deb-etal-2023-zero,
title = "Zero-shot Cross-lingual Transfer With Learned Projections Using Unlabeled Target-Language Data",
author = "Deb, Ujan and
Parab, Ridayesh and
Jyothi, Preethi",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.39",
doi = "10.18653/v1/2023.acl-short.39",
pages = "449--457",
abstract = "Adapters have emerged as a parameter-efficient Transformer-based framework for cross-lingual transfer by inserting lightweight language-specific modules (language adapters) and task-specific modules (task adapters) within pretrained multilingual models. Zero-shot transfer is enabled by pairing the language adapter in the target language with an appropriate task adapter in a source language. If our target languages are known apriori, we explore how zero-shot transfer can be further improved within the adapter framework by utilizing unlabeled text during task-specific finetuning. We construct language-specific subspaces using standard linear algebra constructs and selectively project source-language representations into the target language subspace during task-specific finetuning using two schemes. Our experiments on three cross-lingual tasks, Named Entity Recognition (NER), Question Answering (QA) and Natural Language Inference (NLI) yield consistent benefits compared to adapter baselines over a wide variety of target languages with up to 11{\%} relative improvement in NER, 2{\%} relative improvement in QA and 5{\%} relative improvement in NLI.",
}
| Adapters have emerged as a parameter-efficient Transformer-based framework for cross-lingual transfer by inserting lightweight language-specific modules (language adapters) and task-specific modules (task adapters) within pretrained multilingual models. Zero-shot transfer is enabled by pairing the language adapter in the target language with an appropriate task adapter in a source language. If our target languages are known apriori, we explore how zero-shot transfer can be further improved within the adapter framework by utilizing unlabeled text during task-specific finetuning. We construct language-specific subspaces using standard linear algebra constructs and selectively project source-language representations into the target language subspace during task-specific finetuning using two schemes. Our experiments on three cross-lingual tasks, Named Entity Recognition (NER), Question Answering (QA) and Natural Language Inference (NLI) yield consistent benefits compared to adapter baselines over a wide variety of target languages with up to 11{\%} relative improvement in NER, 2{\%} relative improvement in QA and 5{\%} relative improvement in NLI. | [
"Deb, Ujan",
"Parab, Ridayesh",
"Jyothi, Preethi"
] | Zero-shot Cross-lingual Transfer With Learned Projections Using Unlabeled Target-Language Data | acl-short.39 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.40.bib | https://aclanthology.org/2023.acl-short.40/ | @inproceedings{di-liello-etal-2023-context,
title = "Context-Aware Transformer Pre-Training for Answer Sentence Selection",
author = "Di Liello, Luca and
Garg, Siddhant and
Moschitti, Alessandro",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.40",
doi = "10.18653/v1/2023.acl-short.40",
pages = "458--468",
abstract = "Answer Sentence Selection (AS2) is a core component for building an accurate Question Answering pipeline. AS2 models rank a set of candidate sentences based on how likely they answer a given question. The state of the art in AS2 exploits pre-trained transformers by transferring them on large annotated datasets, while using local contextual information around the candidate sentence. In this paper, we propose three pre-training objectives designed to mimic the downstream fine-tuning task of contextual AS2. This allows for specializing LMs when fine-tuning for contextual AS2. Our experiments on three public and two large-scale industrial datasets show that our pre-training approaches (applied to RoBERTa and ELECTRA) can improve baseline contextual AS2 accuracy by up to 8{\%} on some datasets.",
}
| Answer Sentence Selection (AS2) is a core component for building an accurate Question Answering pipeline. AS2 models rank a set of candidate sentences based on how likely they answer a given question. The state of the art in AS2 exploits pre-trained transformers by transferring them on large annotated datasets, while using local contextual information around the candidate sentence. In this paper, we propose three pre-training objectives designed to mimic the downstream fine-tuning task of contextual AS2. This allows for specializing LMs when fine-tuning for contextual AS2. Our experiments on three public and two large-scale industrial datasets show that our pre-training approaches (applied to RoBERTa and ELECTRA) can improve baseline contextual AS2 accuracy by up to 8{\%} on some datasets. | [
"Di Liello, Luca",
"Garg, Siddhant",
"Moschitti, Aless",
"ro"
] | Context-Aware Transformer Pre-Training for Answer Sentence Selection | acl-short.40 | Poster | 2305.15358 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.41.bib | https://aclanthology.org/2023.acl-short.41/ | @inproceedings{chen-etal-2023-toward,
title = "Toward Expanding the Scope of Radiology Report Summarization to Multiple Anatomies and Modalities",
author = "Chen, Zhihong and
Varma, Maya and
Wan, Xiang and
Langlotz, Curtis and
Delbrouck, Jean-Benoit",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.41",
doi = "10.18653/v1/2023.acl-short.41",
pages = "469--484",
abstract = "Radiology report summarization (RRS) is a growing area of research. Given the Findings section of a radiology report, the goal is to generate a summary (called an Impression section) that highlights the key observations and conclusions of the radiology study. However, RRS currently faces essential limitations. First, many prior studies conduct experiments on private datasets, preventing reproduction of results and fair comparisons across different systems and solutions. Second, most prior approaches are evaluated solely on chest X-rays. To address these limitations, we propose a dataset (MIMIC-RRS) involving three new modalities and seven new anatomies based on the MIMIC-III and MIMIC-CXR datasets. We then conduct extensive experiments to evaluate the performance of models both within and across modality-anatomy pairs in MIMIC-RRS. In addition, we evaluate their clinical efficacy via RadGraph, a factual correctness metric.",
}
| Radiology report summarization (RRS) is a growing area of research. Given the Findings section of a radiology report, the goal is to generate a summary (called an Impression section) that highlights the key observations and conclusions of the radiology study. However, RRS currently faces essential limitations. First, many prior studies conduct experiments on private datasets, preventing reproduction of results and fair comparisons across different systems and solutions. Second, most prior approaches are evaluated solely on chest X-rays. To address these limitations, we propose a dataset (MIMIC-RRS) involving three new modalities and seven new anatomies based on the MIMIC-III and MIMIC-CXR datasets. We then conduct extensive experiments to evaluate the performance of models both within and across modality-anatomy pairs in MIMIC-RRS. In addition, we evaluate their clinical efficacy via RadGraph, a factual correctness metric. | [
"Chen, Zhihong",
"Varma, Maya",
"Wan, Xiang",
"Langlotz, Curtis",
"Delbrouck, Jean-Benoit"
] | Toward Expanding the Scope of Radiology Report Summarization to Multiple Anatomies and Modalities | acl-short.41 | Poster | 2211.08584 | [
"https://github.com/jbdel/vilmedic"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.42.bib | https://aclanthology.org/2023.acl-short.42/ | @inproceedings{blankemeier-etal-2023-efficient,
title = "Efficient Diagnosis Assignment Using Unstructured Clinical Notes",
author = "Blankemeier, Louis and
Fries, Jason and
Tinn, Robert and
Preston, Joseph and
Shah, Nigam and
Chaudhari, Akshay",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.42",
doi = "10.18653/v1/2023.acl-short.42",
pages = "485--494",
abstract = "Electronic phenotyping entails using electronic health records (EHRs) to identify patients with specific health outcomes and determine when those outcomes occurred. Unstructured clinical notes, which contain a vast amount of information, are a valuable resource for electronic phenotyping. However, traditional methods, such as rule-based labeling functions or neural networks, require significant manual effort to tune and may not generalize well to multiple indications. To address these challenges, we propose \textit{HyDE} (hybrid diagnosis extractor). HyDE is a simple framework for electronic phenotyping that integrates labeling functions and a disease-agnostic neural network to assign diagnoses to patients. By training HyDE{'}s model to correct predictions made by labeling functions, we are able to disambiguate hypertension true positives and false positives with a supervised area under the precision-recall curve (AUPRC) of 0.85. We extend this hypertension-trained model to zero-shot evaluation of four other diseases, generating AUPRC values ranging from 0.82 - 0.95 and outperforming a labeling function baseline by 44 points in F1 score and a Word2Vec baseline by 24 points in F1 score on average. Furthermore, we demonstrate a speedup of {\textgreater}4x by pruning the length of inputs into our language model to {\textasciitilde}2.3{\%} of the full clinical notes, with negligible impact to the AUPRC. HyDE has the potential to improve the efficiency and efficacy of interpreting large-scale unstructured clinical notes for accurate EHR phenotyping.",
}
| Electronic phenotyping entails using electronic health records (EHRs) to identify patients with specific health outcomes and determine when those outcomes occurred. Unstructured clinical notes, which contain a vast amount of information, are a valuable resource for electronic phenotyping. However, traditional methods, such as rule-based labeling functions or neural networks, require significant manual effort to tune and may not generalize well to multiple indications. To address these challenges, we propose \textit{HyDE} (hybrid diagnosis extractor). HyDE is a simple framework for electronic phenotyping that integrates labeling functions and a disease-agnostic neural network to assign diagnoses to patients. By training HyDE{'}s model to correct predictions made by labeling functions, we are able to disambiguate hypertension true positives and false positives with a supervised area under the precision-recall curve (AUPRC) of 0.85. We extend this hypertension-trained model to zero-shot evaluation of four other diseases, generating AUPRC values ranging from 0.82 - 0.95 and outperforming a labeling function baseline by 44 points in F1 score and a Word2Vec baseline by 24 points in F1 score on average. Furthermore, we demonstrate a speedup of {\textgreater}4x by pruning the length of inputs into our language model to {\textasciitilde}2.3{\%} of the full clinical notes, with negligible impact to the AUPRC. HyDE has the potential to improve the efficiency and efficacy of interpreting large-scale unstructured clinical notes for accurate EHR phenotyping. | [
"Blankemeier, Louis",
"Fries, Jason",
"Tinn, Robert",
"Preston, Joseph",
"Shah, Nigam",
"Chaudhari, Akshay"
] | Efficient Diagnosis Assignment Using Unstructured Clinical Notes | acl-short.42 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.43.bib | https://aclanthology.org/2023.acl-short.43/ | @inproceedings{monajatipoor-etal-2023-metavl,
title = "{M}eta{VL}: Transferring In-Context Learning Ability From Language Models to Vision-Language Models",
author = "Monajatipoor, Masoud and
Li, Liunian Harold and
Rouhsedaghat, Mozhdeh and
Yang, Lin and
Chang, Kai-Wei",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.43",
doi = "10.18653/v1/2023.acl-short.43",
pages = "495--508",
abstract = "Large-scale language models have shown the ability to adapt to a new task via conditioning on a few demonstrations (i.e., in-context learning). However, in the vision-language domain, most large-scale pre-trained vision-language (VL) models do not possess the ability to conduct in-context learning. How can we enable in-context learning for VL models? In this paper, we study an interesting hypothesis: can we transfer the in-context learning ability from the language domain to the VL domain? Specifically, we first meta-trains a language model to perform in-context learning on NLP tasks (as in MetaICL); then we transfer this model to perform VL tasks by attaching a visual encoder. Our experiments suggest that indeed in-context learning ability can be transferred cross modalities: our model considerably improves the in-context learning capability on VL tasks and can even compensate for the size of the model significantly. On VQA, OK-VQA, and GQA, our method could outperform the baseline model while having {\textasciitilde}20 times fewer parameters.",
}
| Large-scale language models have shown the ability to adapt to a new task via conditioning on a few demonstrations (i.e., in-context learning). However, in the vision-language domain, most large-scale pre-trained vision-language (VL) models do not possess the ability to conduct in-context learning. How can we enable in-context learning for VL models? In this paper, we study an interesting hypothesis: can we transfer the in-context learning ability from the language domain to the VL domain? Specifically, we first meta-trains a language model to perform in-context learning on NLP tasks (as in MetaICL); then we transfer this model to perform VL tasks by attaching a visual encoder. Our experiments suggest that indeed in-context learning ability can be transferred cross modalities: our model considerably improves the in-context learning capability on VL tasks and can even compensate for the size of the model significantly. On VQA, OK-VQA, and GQA, our method could outperform the baseline model while having {\textasciitilde}20 times fewer parameters. | [
"Monajatipoor, Masoud",
"Li, Liunian Harold",
"Rouhsedaghat, Mozhdeh",
"Yang, Lin",
"Chang, Kai-Wei"
] | MetaVL: Transferring In-Context Learning Ability From Language Models to Vision-Language Models | acl-short.43 | Poster | 2306.01311 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.44.bib | https://aclanthology.org/2023.acl-short.44/ | @inproceedings{valentini-etal-2023-interpretability,
title = "On the Interpretability and Significance of Bias Metrics in Texts: a {PMI}-based Approach",
author = "Valentini, Francisco and
Rosati, Germ{\'a}n and
Blasi, Dami{\'a}n and
Fernandez Slezak, Diego and
Altszyler, Edgar",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.44",
doi = "10.18653/v1/2023.acl-short.44",
pages = "509--520",
abstract = "In recent years, word embeddings have been widely used to measure biases in texts. Even if they have proven to be effective in detecting a wide variety of biases, metrics based on word embeddings lack transparency and interpretability. We analyze an alternative PMI-based metric to quantify biases in texts. It can be expressed as a function of conditional probabilities, which provides a simple interpretation in terms of word co-occurrences. We also prove that it can be approximated by an odds ratio, which allows estimating confidence intervals and statistical significance of textual biases. This approach produces similar results to metrics based on word embeddings when capturing gender gaps of the real world embedded in large corpora.",
}
| In recent years, word embeddings have been widely used to measure biases in texts. Even if they have proven to be effective in detecting a wide variety of biases, metrics based on word embeddings lack transparency and interpretability. We analyze an alternative PMI-based metric to quantify biases in texts. It can be expressed as a function of conditional probabilities, which provides a simple interpretation in terms of word co-occurrences. We also prove that it can be approximated by an odds ratio, which allows estimating confidence intervals and statistical significance of textual biases. This approach produces similar results to metrics based on word embeddings when capturing gender gaps of the real world embedded in large corpora. | [
"Valentini, Francisco",
"Rosati, Germ{\\'a}n",
"Blasi, Dami{\\'a}n",
"Fern",
"ez Slezak, Diego",
"Altszyler, Edgar"
] | On the Interpretability and Significance of Bias Metrics in Texts: a PMI-based Approach | acl-short.44 | Poster | 2104.06474 | [
"https://github.com/ftvalentini/biaspmi"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.45.bib | https://aclanthology.org/2023.acl-short.45/ | @inproceedings{doostmohammadi-etal-2023-surface,
title = "Surface-Based Retrieval Reduces Perplexity of Retrieval-Augmented Language Models",
author = "Doostmohammadi, Ehsan and
Norlund, Tobias and
Kuhlmann, Marco and
Johansson, Richard",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.45",
doi = "10.18653/v1/2023.acl-short.45",
pages = "521--529",
abstract = "Augmenting language models with a retrieval mechanism has been shown to significantly improve their performance while keeping the number of parameters low. Retrieval-augmented models commonly rely on a semantic retrieval mechanism based on the similarity between dense representations of the query chunk and potential neighbors. In this paper, we study the state-of-the-art Retro model and observe that its performance gain is better explained by surface-level similarities, such as token overlap. Inspired by this, we replace the semantic retrieval in Retro with a surface-level method based on BM25, obtaining a significant reduction in perplexity. As full BM25 retrieval can be computationally costly for large datasets, we also apply it in a re-ranking scenario, gaining part of the perplexity reduction with minimal computational overhead.",
}
| Augmenting language models with a retrieval mechanism has been shown to significantly improve their performance while keeping the number of parameters low. Retrieval-augmented models commonly rely on a semantic retrieval mechanism based on the similarity between dense representations of the query chunk and potential neighbors. In this paper, we study the state-of-the-art Retro model and observe that its performance gain is better explained by surface-level similarities, such as token overlap. Inspired by this, we replace the semantic retrieval in Retro with a surface-level method based on BM25, obtaining a significant reduction in perplexity. As full BM25 retrieval can be computationally costly for large datasets, we also apply it in a re-ranking scenario, gaining part of the perplexity reduction with minimal computational overhead. | [
"Doostmohammadi, Ehsan",
"Norlund, Tobias",
"Kuhlmann, Marco",
"Johansson, Richard"
] | Surface-Based Retrieval Reduces Perplexity of Retrieval-Augmented Language Models | acl-short.45 | Oral | 2305.16243 | [
"https://github.com/edoost/retro_bm25"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.46.bib | https://aclanthology.org/2023.acl-short.46/ | @inproceedings{razdaibiedina-brechalov-2023-miread,
title = "{MIR}e{AD}: Simple Method for Learning High-quality Representations from Scientific Documents",
author = "Razdaibiedina, Anastasiia and
Brechalov, Aleksandr",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.46",
doi = "10.18653/v1/2023.acl-short.46",
pages = "530--539",
abstract = "Learning semantically meaningful representations from scientific documents can facilitate academic literature search and improve performance of recommendation systems. Pretrained language models have been shown to learn rich textual representations, yet they cannot provide powerful document-level representations for scientific articles. We propose MIReAD, a simple method that learns highquality representations of scientific papers by fine-tuning transformer model to predict the target journal class based on the abstract. We train MIReAD on more than 500,000 PubMed and arXiv abstracts across over 2,000 journal classes. We show that MIReAD produces representations that can be used for similar papers retrieval, topic categorization and literature search. Our proposed approach outperforms six existing models for representation learning on scientific documents across four evaluation standards.",
}
| Learning semantically meaningful representations from scientific documents can facilitate academic literature search and improve performance of recommendation systems. Pretrained language models have been shown to learn rich textual representations, yet they cannot provide powerful document-level representations for scientific articles. We propose MIReAD, a simple method that learns highquality representations of scientific papers by fine-tuning transformer model to predict the target journal class based on the abstract. We train MIReAD on more than 500,000 PubMed and arXiv abstracts across over 2,000 journal classes. We show that MIReAD produces representations that can be used for similar papers retrieval, topic categorization and literature search. Our proposed approach outperforms six existing models for representation learning on scientific documents across four evaluation standards. | [
"Razdaibiedina, Anastasiia",
"Brechalov, Aleks",
"r"
] | MIReAD: Simple Method for Learning High-quality Representations from Scientific Documents | acl-short.46 | Poster | 2305.04177 | [
"https://github.com/arazd/miread"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.47.bib | https://aclanthology.org/2023.acl-short.47/ | @inproceedings{jang-etal-2023-know,
title = "{KNOW} How to Make Up Your Mind! Adversarially Detecting and Alleviating Inconsistencies in Natural Language Explanations",
author = "Jang, Myeongjun and
Majumder, Bodhisattwa Prasad and
McAuley, Julian and
Lukasiewicz, Thomas and
Camburu, Oana-Maria",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.47",
doi = "10.18653/v1/2023.acl-short.47",
pages = "540--553",
abstract = "While recent works have been considerably improving the quality of the natural language explanations (NLEs) generated by a model to justify its predictions, there is very limited research in detecting and alleviating inconsistencies among generated NLEs. In this work, we leverage external knowledge bases to significantly improve on an existing adversarial attack for detecting inconsistent NLEs. We apply our attack to high-performing NLE models and show that models with higher NLE quality do not necessarily generate fewer inconsistencies. Moreover, we propose an off-the-shelf mitigation method to alleviate inconsistencies by grounding the model into external background knowledge. Our method decreases the inconsistencies of previous high-performing NLE models as detected by our attack.",
}
| While recent works have been considerably improving the quality of the natural language explanations (NLEs) generated by a model to justify its predictions, there is very limited research in detecting and alleviating inconsistencies among generated NLEs. In this work, we leverage external knowledge bases to significantly improve on an existing adversarial attack for detecting inconsistent NLEs. We apply our attack to high-performing NLE models and show that models with higher NLE quality do not necessarily generate fewer inconsistencies. Moreover, we propose an off-the-shelf mitigation method to alleviate inconsistencies by grounding the model into external background knowledge. Our method decreases the inconsistencies of previous high-performing NLE models as detected by our attack. | [
"Jang, Myeongjun",
"Majumder, Bodhisattwa Prasad",
"McAuley, Julian",
"Lukasiewicz, Thomas",
"Camburu, Oana-Maria"
] | KNOW How to Make Up Your Mind! Adversarially Detecting and Alleviating Inconsistencies in Natural Language Explanations | acl-short.47 | Oral | 2306.02980 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.48.bib | https://aclanthology.org/2023.acl-short.48/ | @inproceedings{sun-etal-2023-measuring,
title = "Measuring the Effect of Influential Messages on Varying Personas",
author = "Sun, Chenkai and
Li, Jinning and
Chan, Hou Pong and
Zhai, ChengXiang and
Ji, Heng",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.48",
doi = "10.18653/v1/2023.acl-short.48",
pages = "554--562",
abstract = "Predicting how a user responds to news events enables important applications such as allowing intelligent agents or content producers to estimate the effect on different communities and revise unreleased messages to prevent unexpected bad outcomes such as social conflict and moral injury. We present a new task, Response Forecasting on Personas for News Media, to estimate the response a persona (characterizing an individual or a group) might have upon seeing a news message. Compared to the previous efforts which only predict generic comments to news, the proposed task not only introduces personalization in the modeling but also predicts the sentiment polarity and intensity of each response. This enables more accurate and comprehensive inference on the mental state of the persona. Meanwhile, the generated sentiment dimensions make the evaluation and application more reliable. We create the first benchmark dataset, which consists of 13,357 responses to 3,847 news headlines from Twitter. We further evaluate the SOTA neural language models with our dataset. The empirical results suggest that the included persona attributes are helpful for the performance of all response dimensions. Our analysis shows that the best-performing models are capable of predicting responses that are consistent with the personas, and as a byproduct, the task formulation also enables many interesting applications in the analysis of social network groups and their opinions, such as the discovery of extreme opinion groups.",
}
| Predicting how a user responds to news events enables important applications such as allowing intelligent agents or content producers to estimate the effect on different communities and revise unreleased messages to prevent unexpected bad outcomes such as social conflict and moral injury. We present a new task, Response Forecasting on Personas for News Media, to estimate the response a persona (characterizing an individual or a group) might have upon seeing a news message. Compared to the previous efforts which only predict generic comments to news, the proposed task not only introduces personalization in the modeling but also predicts the sentiment polarity and intensity of each response. This enables more accurate and comprehensive inference on the mental state of the persona. Meanwhile, the generated sentiment dimensions make the evaluation and application more reliable. We create the first benchmark dataset, which consists of 13,357 responses to 3,847 news headlines from Twitter. We further evaluate the SOTA neural language models with our dataset. The empirical results suggest that the included persona attributes are helpful for the performance of all response dimensions. Our analysis shows that the best-performing models are capable of predicting responses that are consistent with the personas, and as a byproduct, the task formulation also enables many interesting applications in the analysis of social network groups and their opinions, such as the discovery of extreme opinion groups. | [
"Sun, Chenkai",
"Li, Jinning",
"Chan, Hou Pong",
"Zhai, ChengXiang",
"Ji, Heng"
] | Measuring the Effect of Influential Messages on Varying Personas | acl-short.48 | Poster | 2305.16470 | [
"https://github.com/chenkaisun/response_forecasting"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.49.bib | https://aclanthology.org/2023.acl-short.49/ | @inproceedings{wang-yu-2023-going,
title = "Going Beyond Sentence Embeddings: A Token-Level Matching Algorithm for Calculating Semantic Textual Similarity",
author = "Wang, Hongwei and
Yu, Dong",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.49",
doi = "10.18653/v1/2023.acl-short.49",
pages = "563--570",
abstract = "Semantic Textual Similarity (STS) measures the degree to which the underlying semantics of paired sentences are equivalent. State-of-the-art methods for STS task use language models to encode sentences into embeddings. However, these embeddings are limited in representing semantics because they mix all the semantic information together in fixed-length vectors, which are difficult to recover and lack explainability. This paper presents a token-level matching inference algorithm, which can be applied on top of any language model to improve its performance on STS task. Our method calculates pairwise token-level similarity and token matching scores, and then aggregates them with pretrained token weights to produce sentence similarity. Experimental results on seven STS datasets show that our method improves the performance of almost all language models, with up to 12.7{\%} gain in Spearman{'}s correlation. We also demonstrate that our method is highly explainable and computationally efficient.",
}
| Semantic Textual Similarity (STS) measures the degree to which the underlying semantics of paired sentences are equivalent. State-of-the-art methods for STS task use language models to encode sentences into embeddings. However, these embeddings are limited in representing semantics because they mix all the semantic information together in fixed-length vectors, which are difficult to recover and lack explainability. This paper presents a token-level matching inference algorithm, which can be applied on top of any language model to improve its performance on STS task. Our method calculates pairwise token-level similarity and token matching scores, and then aggregates them with pretrained token weights to produce sentence similarity. Experimental results on seven STS datasets show that our method improves the performance of almost all language models, with up to 12.7{\%} gain in Spearman{'}s correlation. We also demonstrate that our method is highly explainable and computationally efficient. | [
"Wang, Hongwei",
"Yu, Dong"
] | Going Beyond Sentence Embeddings: A Token-Level Matching Algorithm for Calculating Semantic Textual Similarity | acl-short.49 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.50.bib | https://aclanthology.org/2023.acl-short.50/ | @inproceedings{zhu-etal-2023-robust,
title = "Robust Learning for Multi-party Addressee Recognition with Discrete Addressee Codebook",
author = "Zhu, Pengcheng and
Zhou, Wei and
Zhang, Kuncai and
Ma, Yuankai and
Chen, Haiqing",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.50",
doi = "10.18653/v1/2023.acl-short.50",
pages = "571--578",
abstract = "Addressee recognition aims to identify addressees in multi-party conversations. While state-of-the-art addressee recognition models have achieved promising performance, they still suffer from the issue of robustness when applied in real-world scenes. When exposed to a noisy environment, these models regard the noise as input and identify the addressee in a pre-given addressee closed set, while the addressees of the noise do not belong to this closed set, thus leading to the wrong identification of addressee. To this end, we propose a Robust Addressee Recognition (RAR) method, which discrete the addressees into a character codebook, making it able to represent open set addressees and robust in a noisy environment. Experimental results show that the introduction of the addressee character codebook helps to represent the open set addressees and highly improves the robustness of addressee recognition even if the input is noise.",
}
| Addressee recognition aims to identify addressees in multi-party conversations. While state-of-the-art addressee recognition models have achieved promising performance, they still suffer from the issue of robustness when applied in real-world scenes. When exposed to a noisy environment, these models regard the noise as input and identify the addressee in a pre-given addressee closed set, while the addressees of the noise do not belong to this closed set, thus leading to the wrong identification of addressee. To this end, we propose a Robust Addressee Recognition (RAR) method, which discrete the addressees into a character codebook, making it able to represent open set addressees and robust in a noisy environment. Experimental results show that the introduction of the addressee character codebook helps to represent the open set addressees and highly improves the robustness of addressee recognition even if the input is noise. | [
"Zhu, Pengcheng",
"Zhou, Wei",
"Zhang, Kuncai",
"Ma, Yuankai",
"Chen, Haiqing"
] | Robust Learning for Multi-party Addressee Recognition with Discrete Addressee Codebook | acl-short.50 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.51.bib | https://aclanthology.org/2023.acl-short.51/ | @inproceedings{loakman-etal-2023-twistlist,
title = "{T}wist{L}ist: Resources and Baselines for Tongue Twister Generation",
author = "Loakman, Tyler and
Tang, Chen and
Lin, Chenghua",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.51",
doi = "10.18653/v1/2023.acl-short.51",
pages = "579--589",
abstract = "Previous work in phonetically-grounded language generation has mainly focused on domains such as lyrics and poetry. In this paper, we present work on the generation of tongue twisters - a form of language that is required to be phonetically conditioned to maximise sound overlap, whilst maintaining semantic consistency with an input topic, and still being grammatically correct. We present TwistList, a large annotated dataset of tongue twisters, consisting of 2.1K+ human-authored examples. We additionally present several benchmark systems (referred to as TwisterMisters) for the proposed task of tongue twister generation, including models that both do and do not require training on in-domain data. We present the results of automatic and human evaluation to demonstrate the performance ofexisting mainstream pre-trained models in this task with limited (or no) task specific training and data, and no explicit phonetic knowledge. We find that the task of tongue twister generation is challenging for models under these conditions, yet some models are still capable of generating acceptable examples of this language type.",
}
| Previous work in phonetically-grounded language generation has mainly focused on domains such as lyrics and poetry. In this paper, we present work on the generation of tongue twisters - a form of language that is required to be phonetically conditioned to maximise sound overlap, whilst maintaining semantic consistency with an input topic, and still being grammatically correct. We present TwistList, a large annotated dataset of tongue twisters, consisting of 2.1K+ human-authored examples. We additionally present several benchmark systems (referred to as TwisterMisters) for the proposed task of tongue twister generation, including models that both do and do not require training on in-domain data. We present the results of automatic and human evaluation to demonstrate the performance ofexisting mainstream pre-trained models in this task with limited (or no) task specific training and data, and no explicit phonetic knowledge. We find that the task of tongue twister generation is challenging for models under these conditions, yet some models are still capable of generating acceptable examples of this language type. | [
"Loakman, Tyler",
"Tang, Chen",
"Lin, Chenghua"
] | TwistList: Resources and Baselines for Tongue Twister Generation | acl-short.51 | Poster | 2306.03457 | [
"https://github.com/tangg555/twistlist"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.52.bib | https://aclanthology.org/2023.acl-short.52/ | @inproceedings{card-2023-substitution,
title = "Substitution-based Semantic Change Detection using Contextual Embeddings",
author = "Card, Dallas",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.52",
doi = "10.18653/v1/2023.acl-short.52",
pages = "590--602",
abstract = "Measuring semantic change has thus far remained a task where methods using contextual embeddings have struggled to improve upon simpler techniques relying only on static word vectors. Moreover, many of the previously proposed approaches suffer from downsides related to scalability and ease of interpretation. We present a simplified approach to measuring semantic change using contextual embeddings, relying only on the most probable substitutes for masked terms. Not only is this approach directly interpretable, it is also far more efficient in terms of storage, achieves superior average performance across the most frequently cited datasets for this task, and allows for more nuanced investigation of change than is possible with static word vectors.",
}
| Measuring semantic change has thus far remained a task where methods using contextual embeddings have struggled to improve upon simpler techniques relying only on static word vectors. Moreover, many of the previously proposed approaches suffer from downsides related to scalability and ease of interpretation. We present a simplified approach to measuring semantic change using contextual embeddings, relying only on the most probable substitutes for masked terms. Not only is this approach directly interpretable, it is also far more efficient in terms of storage, achieves superior average performance across the most frequently cited datasets for this task, and allows for more nuanced investigation of change than is possible with static word vectors. | [
"Card, Dallas"
] | Substitution-based Semantic Change Detection using Contextual Embeddings | acl-short.52 | Poster | 2309.02403 | [
"https://github.com/dallascard/SBSCD"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.53.bib | https://aclanthology.org/2023.acl-short.53/ | @inproceedings{kondo-etal-2023-probing,
title = "Probing Physical Reasoning with Counter-Commonsense Context",
author = "Kondo, Kazushi and
Sugawara, Saku and
Aizawa, Akiko",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.53",
doi = "10.18653/v1/2023.acl-short.53",
pages = "603--612",
abstract = "In this study, we create a CConS (Counter-commonsense Contextual Size comparison) dataset to investigate how physical commonsense affects the contextualized size comparison task; the proposed dataset consists of both contexts that fit physical commonsense and those that do not. This dataset tests the ability of language models to predict the size relationship between objects under various contexts generated from our curated noun list and templates. We measure the ability of several masked language models and encoder-decoder models. The results show that while large language models can use prepositions such as {``}in{''} and {``}into{''} in the provided context to infer size relationships, they fail to use verbs and thus make incorrect judgments led by their prior physical commonsense.",
}
| In this study, we create a CConS (Counter-commonsense Contextual Size comparison) dataset to investigate how physical commonsense affects the contextualized size comparison task; the proposed dataset consists of both contexts that fit physical commonsense and those that do not. This dataset tests the ability of language models to predict the size relationship between objects under various contexts generated from our curated noun list and templates. We measure the ability of several masked language models and encoder-decoder models. The results show that while large language models can use prepositions such as {``}in{''} and {``}into{''} in the provided context to infer size relationships, they fail to use verbs and thus make incorrect judgments led by their prior physical commonsense. | [
"Kondo, Kazushi",
"Sugawara, Saku",
"Aizawa, Akiko"
] | Probing Physical Reasoning with Counter-Commonsense Context | acl-short.53 | Poster | 2306.02258 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.54.bib | https://aclanthology.org/2023.acl-short.54/ | @inproceedings{guriel-etal-2023-morphological,
title = "Morphological Inflection with Phonological Features",
author = "Guriel, David and
Goldman, Omer and
Tsarfaty, Reut",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.54",
doi = "10.18653/v1/2023.acl-short.54",
pages = "613--622",
abstract = "Recent years have brought great advances into solving morphological tasks, mostly due to powerful neural models applied to various tasks as (re)inflection and analysis. Yet, such morphological tasks cannot be considered solved, especially when little training data is available or when generalizing to previously unseen lemmas. This work explores effects on performance obtained through various ways in which morphological models get access to sub-character phonological features that are often the targets of morphological processes. We design two methods to achieve this goal: one that leaves models as is but manipulates the data to include features instead of characters, and another that manipulates models to take phonological features into account when building representations for phonemes. We elicit phonemic data from standard graphemic data using language-specific grammars for languages with shallow grapheme-to-phoneme mapping, and we experiment with two reinflection models over eight languages. Our results show that our methods yield comparable results to the grapheme-based baseline overall, with minor improvements in some of the languages. All in all, we conclude that patterns in character distributions are likely to allow models to infer the underlying phonological characteristics, even when phonemes are not explicitly represented.",
}
| Recent years have brought great advances into solving morphological tasks, mostly due to powerful neural models applied to various tasks as (re)inflection and analysis. Yet, such morphological tasks cannot be considered solved, especially when little training data is available or when generalizing to previously unseen lemmas. This work explores effects on performance obtained through various ways in which morphological models get access to sub-character phonological features that are often the targets of morphological processes. We design two methods to achieve this goal: one that leaves models as is but manipulates the data to include features instead of characters, and another that manipulates models to take phonological features into account when building representations for phonemes. We elicit phonemic data from standard graphemic data using language-specific grammars for languages with shallow grapheme-to-phoneme mapping, and we experiment with two reinflection models over eight languages. Our results show that our methods yield comparable results to the grapheme-based baseline overall, with minor improvements in some of the languages. All in all, we conclude that patterns in character distributions are likely to allow models to infer the underlying phonological characteristics, even when phonemes are not explicitly represented. | [
"Guriel, David",
"Goldman, Omer",
"Tsarfaty, Reut"
] | Morphological Inflection with Phonological Features | acl-short.54 | Poster | 2306.12581 | [
"https://github.com/onlplab/inflectionwithphonology"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.55.bib | https://aclanthology.org/2023.acl-short.55/ | @inproceedings{wu-etal-2023-holistic,
title = "A Holistic Approach to Reference-Free Evaluation of Machine Translation",
author = "Wu, Hanming and
Han, Wenjuan and
Di, Hui and
Chen, Yufeng and
Xu, Jinan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.55",
doi = "10.18653/v1/2023.acl-short.55",
pages = "623--636",
abstract = "Traditional machine translation evaluation relies on reference written by humans. While reference-free evaluation gets rid of the constraints of labor-intensive annotations, which can pivot easily to new domains and is more scalable. In this paper, we propose a reference-free evaluation approach that characterizes evaluation as two aspects: (1) fluency: how well the translated text conforms to normal human language usage; (2) faithfulness: how well the translated text reflects the source data. We further split the faithfulness into word-level and sentence-level. Extensive experiments spanning WMT18/19/21 Metrics segment-level daRR and MQM datasets demonstrate that our proposed reference-free approach, ReFreeEval, outperforms SOTA reference-fee metrics like YiSi-2.",
}
| Traditional machine translation evaluation relies on reference written by humans. While reference-free evaluation gets rid of the constraints of labor-intensive annotations, which can pivot easily to new domains and is more scalable. In this paper, we propose a reference-free evaluation approach that characterizes evaluation as two aspects: (1) fluency: how well the translated text conforms to normal human language usage; (2) faithfulness: how well the translated text reflects the source data. We further split the faithfulness into word-level and sentence-level. Extensive experiments spanning WMT18/19/21 Metrics segment-level daRR and MQM datasets demonstrate that our proposed reference-free approach, ReFreeEval, outperforms SOTA reference-fee metrics like YiSi-2. | [
"Wu, Hanming",
"Han, Wenjuan",
"Di, Hui",
"Chen, Yufeng",
"Xu, Jinan"
] | A Holistic Approach to Reference-Free Evaluation of Machine Translation | acl-short.55 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.56.bib | https://aclanthology.org/2023.acl-short.56/ | @inproceedings{sul-choi-2023-balancing,
title = "Balancing Lexical and Semantic Quality in Abstractive Summarization",
author = "Sul, Jeewoo and
Choi, Yong Suk",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.56",
doi = "10.18653/v1/2023.acl-short.56",
pages = "637--647",
abstract = "An important problem of the sequence-to-sequence neural models widely used in abstractive summarization is exposure bias. To alleviate this problem, re-ranking systems have been applied in recent years. Despite some performance improvements, this approach remains underexplored. Previous works have mostly specified the rank through the ROUGE score and aligned candidate summaries, but there can be quite a large gap between the lexical overlap metric and semantic similarity. In this paper, we propose a novel training method in which a re-ranker balances the lexical and semantic quality. We further newly define false positives in ranking and present a strategy to reduce their influence. Experiments on the CNN/DailyMail and XSum datasets show that our method can estimate the meaning of summaries without seriously degrading the lexical aspect. More specifically, it achieves an 89.67 BERTScore on the CNN/DailyMail dataset, reaching new state-of-the-art performance. Our code is publicly available at \url{https://github.com/jeewoo1025/BalSum}.",
}
| An important problem of the sequence-to-sequence neural models widely used in abstractive summarization is exposure bias. To alleviate this problem, re-ranking systems have been applied in recent years. Despite some performance improvements, this approach remains underexplored. Previous works have mostly specified the rank through the ROUGE score and aligned candidate summaries, but there can be quite a large gap between the lexical overlap metric and semantic similarity. In this paper, we propose a novel training method in which a re-ranker balances the lexical and semantic quality. We further newly define false positives in ranking and present a strategy to reduce their influence. Experiments on the CNN/DailyMail and XSum datasets show that our method can estimate the meaning of summaries without seriously degrading the lexical aspect. More specifically, it achieves an 89.67 BERTScore on the CNN/DailyMail dataset, reaching new state-of-the-art performance. Our code is publicly available at \url{https://github.com/jeewoo1025/BalSum}. | [
"Sul, Jeewoo",
"Choi, Yong Suk"
] | Balancing Lexical and Semantic Quality in Abstractive Summarization | acl-short.56 | Oral | 2305.09898 | [
"https://github.com/jeewoo1025/balsum"
] | https://huggingface.co/papers/2305.09898 | 0 | 1 | 0 | 2 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-short.57.bib | https://aclanthology.org/2023.acl-short.57/ | @inproceedings{agravante-etal-2023-learning,
title = "Learning Neuro-Symbolic World Models with Conversational Proprioception",
author = "Agravante, Don Joven and
Kimura, Daiki and
Tatsubori, Michiaki and
Munawar, Asim and
Gray, Alexander",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.57",
doi = "10.18653/v1/2023.acl-short.57",
pages = "648--656",
abstract = "The recent emergence of Neuro-Symbolic Agent (NeSA) approaches to natural language-based interactions calls for the investigation of model-based approaches. In contrast to model-free approaches, which existing NeSAs take, learning an explicit world model has an interesting potential especially in the explainability, which is one of the key selling points of NeSA. To learn useful world models, we leverage one of the recent neuro-symbolic architectures, Logical Neural Networks (LNN). Here, we describe a method that can learn neuro-symbolic world models on the TextWorld-Commonsense set of games. We then show how this can be improved further by taking inspiration from the concept of proprioception, but for conversation. This is done by enhancing the internal logic state with a memory of previous actions while also guiding future actions by augmenting the learned model with constraints based on this memory. This greatly improves the game-solving agents performance in a TextWorld setting, where the advantage over the baseline is an 85{\%} average steps reduction and x2.3 average score.",
}
| The recent emergence of Neuro-Symbolic Agent (NeSA) approaches to natural language-based interactions calls for the investigation of model-based approaches. In contrast to model-free approaches, which existing NeSAs take, learning an explicit world model has an interesting potential especially in the explainability, which is one of the key selling points of NeSA. To learn useful world models, we leverage one of the recent neuro-symbolic architectures, Logical Neural Networks (LNN). Here, we describe a method that can learn neuro-symbolic world models on the TextWorld-Commonsense set of games. We then show how this can be improved further by taking inspiration from the concept of proprioception, but for conversation. This is done by enhancing the internal logic state with a memory of previous actions while also guiding future actions by augmenting the learned model with constraints based on this memory. This greatly improves the game-solving agents performance in a TextWorld setting, where the advantage over the baseline is an 85{\%} average steps reduction and x2.3 average score. | [
"Agravante, Don Joven",
"Kimura, Daiki",
"Tatsubori, Michiaki",
"Munawar, Asim",
"Gray, Alex",
"er"
] | Learning Neuro-Symbolic World Models with Conversational Proprioception | acl-short.57 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.58.bib | https://aclanthology.org/2023.acl-short.58/ | @inproceedings{yang-etal-2023-domain,
title = "In and Out-of-Domain Text Adversarial Robustness via Label Smoothing",
author = "Yang, Yahan and
Dan, Soham and
Roth, Dan and
Lee, Insup",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.58",
doi = "10.18653/v1/2023.acl-short.58",
pages = "657--669",
abstract = "Recently it has been shown that state-of-the-art NLP models are vulnerable to adversarial attacks, where the predictions of a model can be drastically altered by slight modifications to the input (such as synonym substitutions). While several defense techniques have been proposed, and adapted, to the discrete nature of text adversarial attacks, the benefits of general-purpose regularization methods such as label smoothing for language models, have not been studied. In this paper, we study the adversarial robustness provided by label smoothing strategies in foundational models for diverse NLP tasks in both in-domain and out-of-domain settings. Our experiments show that label smoothing significantly improves adversarial robustness in pre-trained models like BERT, against various popular attacks. We also analyze the relationship between prediction confidence and robustness, showing that label smoothing reduces over-confident errors on adversarial examples.",
}
| Recently it has been shown that state-of-the-art NLP models are vulnerable to adversarial attacks, where the predictions of a model can be drastically altered by slight modifications to the input (such as synonym substitutions). While several defense techniques have been proposed, and adapted, to the discrete nature of text adversarial attacks, the benefits of general-purpose regularization methods such as label smoothing for language models, have not been studied. In this paper, we study the adversarial robustness provided by label smoothing strategies in foundational models for diverse NLP tasks in both in-domain and out-of-domain settings. Our experiments show that label smoothing significantly improves adversarial robustness in pre-trained models like BERT, against various popular attacks. We also analyze the relationship between prediction confidence and robustness, showing that label smoothing reduces over-confident errors on adversarial examples. | [
"Yang, Yahan",
"Dan, Soham",
"Roth, Dan",
"Lee, Insup"
] | In and Out-of-Domain Text Adversarial Robustness via Label Smoothing | acl-short.58 | Poster | 2212.10258 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.59.bib | https://aclanthology.org/2023.acl-short.59/ | @inproceedings{abaskohi-etal-2023-lm,
title = "{LM}-{CPPF}: Paraphrasing-Guided Data Augmentation for Contrastive Prompt-Based Few-Shot Fine-Tuning",
author = "Abaskohi, Amirhossein and
Rothe, Sascha and
Yaghoobzadeh, Yadollah",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.59",
doi = "10.18653/v1/2023.acl-short.59",
pages = "670--681",
abstract = "In recent years, there has been significant progress in developing pre-trained language models for NLP. However, these models often struggle when fine-tuned on small datasets. To address this issue, researchers have proposed various adaptation approaches. Prompt-based tuning is arguably the most common way, especially for larger models. Previous research shows that adding contrastive learning to prompt-based fine-tuning is effective as it helps the model generate embeddings that are more distinguishable between classes, and it can also be more sample-efficient as the model learns from positive and negative examples simultaneously. One of the most important components of contrastive learning is data augmentation, but unlike computer vision, effective data augmentation for NLP is still challenging. This paper proposes LM-CPPF, Contrastive Paraphrasing-guided Prompt-based Fine-tuning of Language Models, which leverages prompt-based few-shot paraphrasing using generative language models, especially large language models such as GPT-3 and OPT-175B, for data augmentation. Our experiments on multiple text classification benchmarks show that this augmentation method outperforms other methods, such as easy data augmentation, back translation, and multiple templates.",
}
| In recent years, there has been significant progress in developing pre-trained language models for NLP. However, these models often struggle when fine-tuned on small datasets. To address this issue, researchers have proposed various adaptation approaches. Prompt-based tuning is arguably the most common way, especially for larger models. Previous research shows that adding contrastive learning to prompt-based fine-tuning is effective as it helps the model generate embeddings that are more distinguishable between classes, and it can also be more sample-efficient as the model learns from positive and negative examples simultaneously. One of the most important components of contrastive learning is data augmentation, but unlike computer vision, effective data augmentation for NLP is still challenging. This paper proposes LM-CPPF, Contrastive Paraphrasing-guided Prompt-based Fine-tuning of Language Models, which leverages prompt-based few-shot paraphrasing using generative language models, especially large language models such as GPT-3 and OPT-175B, for data augmentation. Our experiments on multiple text classification benchmarks show that this augmentation method outperforms other methods, such as easy data augmentation, back translation, and multiple templates. | [
"Abaskohi, Amirhossein",
"Rothe, Sascha",
"Yaghoobzadeh, Yadollah"
] | LM-CPPF: Paraphrasing-Guided Data Augmentation for Contrastive Prompt-Based Few-Shot Fine-Tuning | acl-short.59 | Poster | 2305.18169 | [
"https://github.com/amirabaskohi/lm-cppf"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.60.bib | https://aclanthology.org/2023.acl-short.60/ | @inproceedings{muller-etal-2023-considerations,
title = "Considerations for meaningful sign language machine translation based on glosses",
author = {M{\"u}ller, Mathias and
Jiang, Zifan and
Moryossef, Amit and
Rios, Annette and
Ebling, Sarah},
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.60",
doi = "10.18653/v1/2023.acl-short.60",
pages = "682--693",
abstract = "Automatic sign language processing is gaining popularity in Natural Language Processing (NLP) research (Yin et al., 2021). In machine translation (MT) in particular, sign language translation based on glosses is a prominent approach. In this paper, we review recent works on neural gloss translation. We find that limitations of glosses in general and limitations of specific datasets are not discussed in a transparent manner and that there is no common standard for evaluation. To address these issues, we put forward concrete recommendations for future research on gloss translation. Our suggestions advocate awareness of the inherent limitations of gloss-based approaches, realistic datasets, stronger baselines and convincing evaluation.",
}
| Automatic sign language processing is gaining popularity in Natural Language Processing (NLP) research (Yin et al., 2021). In machine translation (MT) in particular, sign language translation based on glosses is a prominent approach. In this paper, we review recent works on neural gloss translation. We find that limitations of glosses in general and limitations of specific datasets are not discussed in a transparent manner and that there is no common standard for evaluation. To address these issues, we put forward concrete recommendations for future research on gloss translation. Our suggestions advocate awareness of the inherent limitations of gloss-based approaches, realistic datasets, stronger baselines and convincing evaluation. | [
"M{\\\"u}ller, Mathias",
"Jiang, Zifan",
"Moryossef, Amit",
"Rios, Annette",
"Ebling, Sarah"
] | Considerations for meaningful sign language machine translation based on glosses | acl-short.60 | Poster | 2211.15464 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.61.bib | https://aclanthology.org/2023.acl-short.61/ | @inproceedings{sosa-etal-2023-detecting,
title = "Detecting Contradictory {COVID}-19 Drug Efficacy Claims from Biomedical Literature",
author = "Sosa, Daniel and
Suresh, Malavika and
Potts, Christopher and
Altman, Russ",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.61",
doi = "10.18653/v1/2023.acl-short.61",
pages = "694--713",
abstract = "The COVID-19 pandemic created a deluge of questionable and contradictory scientific claims about drug efficacy {--} an {``}infodemic{''} with lasting consequences for science and society. In this work, we argue that NLP models can help domain experts distill and understand the literature in this complex, high-stakes area. Our task is to automatically identify contradictory claims about COVID-19 drug efficacy. We frame this as a natural language inference problem and offer a new NLI dataset created by domain experts. The NLI framing allows us to create curricula combining existing datasets and our own. The resulting models are useful investigative tools. We provide a case study of how these models help a domain expert summarize and assess evidence concerning remdisivir and hydroxychloroquine.",
}
| The COVID-19 pandemic created a deluge of questionable and contradictory scientific claims about drug efficacy {--} an {``}infodemic{''} with lasting consequences for science and society. In this work, we argue that NLP models can help domain experts distill and understand the literature in this complex, high-stakes area. Our task is to automatically identify contradictory claims about COVID-19 drug efficacy. We frame this as a natural language inference problem and offer a new NLI dataset created by domain experts. The NLI framing allows us to create curricula combining existing datasets and our own. The resulting models are useful investigative tools. We provide a case study of how these models help a domain expert summarize and assess evidence concerning remdisivir and hydroxychloroquine. | [
"Sosa, Daniel",
"Suresh, Malavika",
"Potts, Christopher",
"Altman, Russ"
] | Detecting Contradictory COVID-19 Drug Efficacy Claims from Biomedical Literature | acl-short.61 | Poster | 2212.09867 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.62.bib | https://aclanthology.org/2023.acl-short.62/ | @inproceedings{amalvy-etal-2023-role,
title = "The Role of Global and Local Context in Named Entity Recognition",
author = "Amalvy, Arthur and
Labatut, Vincent and
Dufour, Richard",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.62",
doi = "10.18653/v1/2023.acl-short.62",
pages = "714--722",
abstract = "Pre-trained transformer-based models have recently shown great performance when applied to Named Entity Recognition (NER). As the complexity of their self-attention mechanism prevents them from processing long documents at once, these models are usually applied in a sequential fashion. Such an approach unfortunately only incorporates local context and prevents leveraging global document context in long documents such as novels, which might hinder performance. In this article, we explore the impact of global document context, and its relationships with local context. We find that correctly retrieving global document context has a greater impact on performance than only leveraging local context, prompting for further research on how to better retrieve that context.",
}
| Pre-trained transformer-based models have recently shown great performance when applied to Named Entity Recognition (NER). As the complexity of their self-attention mechanism prevents them from processing long documents at once, these models are usually applied in a sequential fashion. Such an approach unfortunately only incorporates local context and prevents leveraging global document context in long documents such as novels, which might hinder performance. In this article, we explore the impact of global document context, and its relationships with local context. We find that correctly retrieving global document context has a greater impact on performance than only leveraging local context, prompting for further research on how to better retrieve that context. | [
"Amalvy, Arthur",
"Labatut, Vincent",
"Dufour, Richard"
] | The Role of Global and Local Context in Named Entity Recognition | acl-short.62 | Poster | 2305.03132 | [
"https://github.com/CompNet/conivel"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.63.bib | https://aclanthology.org/2023.acl-short.63/ | @inproceedings{spaulding-etal-2023-joint,
title = "Joint End-to-end Semantic Proto-role Labeling",
author = "Spaulding, Elizabeth and
Kazantsev, Gary and
Dredze, Mark",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.63",
doi = "10.18653/v1/2023.acl-short.63",
pages = "723--736",
abstract = "Semantic proto-role labeling (SPRL) assigns properties to arguments based on a series of binary labels. While multiple studies have evaluated various approaches to SPRL, it has only been studied in-depth as a standalone task using gold predicate/argument pairs. How do SPRL systems perform as part of an information extraction pipeline? We model SPRL jointly with predicate-argument extraction using a deep transformer model. We find that proto-role labeling is surprisingly robust in this setting, with only a small decrease when using predicted arguments. We include a detailed analysis of each component of the joint system, and an error analysis to understand correlations in errors between system stages. Finally, we study the effects of annotation errors on SPRL.",
}
| Semantic proto-role labeling (SPRL) assigns properties to arguments based on a series of binary labels. While multiple studies have evaluated various approaches to SPRL, it has only been studied in-depth as a standalone task using gold predicate/argument pairs. How do SPRL systems perform as part of an information extraction pipeline? We model SPRL jointly with predicate-argument extraction using a deep transformer model. We find that proto-role labeling is surprisingly robust in this setting, with only a small decrease when using predicted arguments. We include a detailed analysis of each component of the joint system, and an error analysis to understand correlations in errors between system stages. Finally, we study the effects of annotation errors on SPRL. | [
"Spaulding, Elizabeth",
"Kazantsev, Gary",
"Dredze, Mark"
] | Joint End-to-end Semantic Proto-role Labeling | acl-short.63 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.64.bib | https://aclanthology.org/2023.acl-short.64/ | @inproceedings{vishnubhotla-etal-2023-improving,
title = "Improving Automatic Quotation Attribution in Literary Novels",
author = "Vishnubhotla, Krishnapriya and
Rudzicz, Frank and
Hirst, Graeme and
Hammond, Adam",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.64",
doi = "10.18653/v1/2023.acl-short.64",
pages = "737--746",
abstract = "Current models for quotation attribution in literary novels assume varying levels of available information in their training and test data, which poses a challenge for in-the-wild inference. Here, we approach quotation attribution as a set of four interconnected sub-tasks: character identification, coreference resolution, quotation identification, and speaker attribution. We benchmark state-of-the-art models on each of these sub-tasks independently, using a large dataset of annotated coreferences and quotations in literary novels (the Project Dialogism Novel Corpus). We also train and evaluate models for the speaker attribution task in particular, showing that a simple sequential prediction model achieves accuracy scores on par with state-of-the-art models.",
}
| Current models for quotation attribution in literary novels assume varying levels of available information in their training and test data, which poses a challenge for in-the-wild inference. Here, we approach quotation attribution as a set of four interconnected sub-tasks: character identification, coreference resolution, quotation identification, and speaker attribution. We benchmark state-of-the-art models on each of these sub-tasks independently, using a large dataset of annotated coreferences and quotations in literary novels (the Project Dialogism Novel Corpus). We also train and evaluate models for the speaker attribution task in particular, showing that a simple sequential prediction model achieves accuracy scores on par with state-of-the-art models. | [
"Vishnubhotla, Krishnapriya",
"Rudzicz, Frank",
"Hirst, Graeme",
"Hammond, Adam"
] | Improving Automatic Quotation Attribution in Literary Novels | acl-short.64 | Poster | 2307.03734 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.65.bib | https://aclanthology.org/2023.acl-short.65/ | @inproceedings{subramanian-etal-2023-modular,
title = "Modular Visual Question Answering via Code Generation",
author = "Subramanian, Sanjay and
Narasimhan, Medhini and
Khangaonkar, Kushal and
Yang, Kevin and
Nagrani, Arsha and
Schmid, Cordelia and
Zeng, Andy and
Darrell, Trevor and
Klein, Dan",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.65",
doi = "10.18653/v1/2023.acl-short.65",
pages = "747--761",
abstract = "We present a framework that formulates visual question answering as modular code generation. In contrast to prior work on modular approaches to VQA, our approach requires no additional training and relies on pre-trained language models (LMs), visual models pre-trained on image-caption pairs, and fifty VQA examples used for in-context learning. The generated Python programs invoke and compose the outputs of the visual models using arithmetic and conditional logic. Our approach improves accuracy on the COVR dataset by at least 3{\%} and on the GQA dataset by 2{\%} compared to the few-shot baseline that does not employ code generation.",
}
| We present a framework that formulates visual question answering as modular code generation. In contrast to prior work on modular approaches to VQA, our approach requires no additional training and relies on pre-trained language models (LMs), visual models pre-trained on image-caption pairs, and fifty VQA examples used for in-context learning. The generated Python programs invoke and compose the outputs of the visual models using arithmetic and conditional logic. Our approach improves accuracy on the COVR dataset by at least 3{\%} and on the GQA dataset by 2{\%} compared to the few-shot baseline that does not employ code generation. | [
"Subramanian, Sanjay",
"Narasimhan, Medhini",
"Khangaonkar, Kushal",
"Yang, Kevin",
"Nagrani, Arsha",
"Schmid, Cordelia",
"Zeng, Andy",
"Darrell, Trevor",
"Klein, Dan"
] | Modular Visual Question Answering via Code Generation | acl-short.65 | Poster | 2306.05392 | [
"https://github.com/sanjayss34/codevqa"
] | https://huggingface.co/papers/2306.05392 | 4 | 2 | 0 | 9 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-short.66.bib | https://aclanthology.org/2023.acl-short.66/ | @inproceedings{zampieri-etal-2023-target,
title = "Target-Based Offensive Language Identification",
author = "Zampieri, Marcos and
Morgan, Skye and
North, Kai and
Ranasinghe, Tharindu and
Simmmons, Austin and
Khandelwal, Paridhi and
Rosenthal, Sara and
Nakov, Preslav",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.66",
doi = "10.18653/v1/2023.acl-short.66",
pages = "762--770",
abstract = "We present TBO, a new dataset for Target-based Offensive language identification. TBO contains post-level annotations regarding the harmfulness of an offensive post and token-level annotations comprising of the target and the offensive argument expression. Popular offensive language identification datasets for social media focus on annotation taxonomies only at the post level and more recently, some datasets have been released that feature only token-level annotations. TBO is an important resource that bridges the gap between post-level and token-level annotation datasets by introducing a single comprehensive unified annotation taxonomy. We use the TBO taxonomy to annotate post-level and token-level offensive language on English Twitter posts. We release an initial dataset of over 4,500 instances collected from Twitter and we carry out multiple experiments to compare the performance of different models trained and tested on TBO.",
}
| We present TBO, a new dataset for Target-based Offensive language identification. TBO contains post-level annotations regarding the harmfulness of an offensive post and token-level annotations comprising of the target and the offensive argument expression. Popular offensive language identification datasets for social media focus on annotation taxonomies only at the post level and more recently, some datasets have been released that feature only token-level annotations. TBO is an important resource that bridges the gap between post-level and token-level annotation datasets by introducing a single comprehensive unified annotation taxonomy. We use the TBO taxonomy to annotate post-level and token-level offensive language on English Twitter posts. We release an initial dataset of over 4,500 instances collected from Twitter and we carry out multiple experiments to compare the performance of different models trained and tested on TBO. | [
"Zampieri, Marcos",
"Morgan, Skye",
"North, Kai",
"Ranasinghe, Tharindu",
"Simmmons, Austin",
"Kh",
"elwal, Paridhi",
"Rosenthal, Sara",
"Nakov, Preslav"
] | Target-Based Offensive Language Identification | acl-short.66 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.67.bib | https://aclanthology.org/2023.acl-short.67/ | @inproceedings{ponce-etal-2023-unsupervised,
title = "Unsupervised Subtitle Segmentation with Masked Language Models",
author = "Ponce, David and
Etchegoyhen, Thierry and
Ruiz, Victor",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.67",
doi = "10.18653/v1/2023.acl-short.67",
pages = "771--781",
abstract = "We describe a novel unsupervised approach to subtitle segmentation, based on pretrained masked language models, where line endings and subtitle breaks are predicted according to the likelihood of punctuation to occur at candidate segmentation points. Our approach obtained competitive results in terms of segmentation accuracy across metrics, while also fully preserving the original text and complying with length constraints. Although supervised models trained on in-domain data and with access to source audio information can provide better segmentation accuracy, our approach is highly portable across languages and domains and may constitute a robust off-the-shelf solution for subtitle segmentation.",
}
| We describe a novel unsupervised approach to subtitle segmentation, based on pretrained masked language models, where line endings and subtitle breaks are predicted according to the likelihood of punctuation to occur at candidate segmentation points. Our approach obtained competitive results in terms of segmentation accuracy across metrics, while also fully preserving the original text and complying with length constraints. Although supervised models trained on in-domain data and with access to source audio information can provide better segmentation accuracy, our approach is highly portable across languages and domains and may constitute a robust off-the-shelf solution for subtitle segmentation. | [
"Ponce, David",
"Etchegoyhen, Thierry",
"Ruiz, Victor"
] | Unsupervised Subtitle Segmentation with Masked Language Models | acl-short.67 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.68.bib | https://aclanthology.org/2023.acl-short.68/ | @inproceedings{yadav-etal-2023-exploring,
title = "Exploring Continual Learning for Code Generation Models",
author = "Yadav, Prateek and
Sun, Qing and
Ding, Hantian and
Li, Xiaopeng and
Zhang, Dejiao and
Tan, Ming and
Bhatia, Parminder and
Ma, Xiaofei and
Nallapati, Ramesh and
Ramanathan, Murali Krishna and
Bansal, Mohit and
Xiang, Bing",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.68",
doi = "10.18653/v1/2023.acl-short.68",
pages = "782--792",
abstract = "Large-scale code generation models such as Copilot and CodeT5 have achieved impressive performance. However, libraries are upgraded or deprecated very frequently and re-training large-scale language models is computationally expensive. Therefore, Continual Learning (CL) is an important aspect that remains under-explored in the code domain. In this paper, we introduce a benchmark called CodeTask-CL that covers a wide range of tasks, including code generation, translation, summarization, and refinement, with different input and output programming languages. Next, on our CodeTask-CL benchmark, we compare popular CL techniques from NLP and Vision domains. We find that effective methods like Prompt Pooling (PP) suffer from catastrophic forgetting due to the unstable training of the prompt selection mechanism caused by stark distribution shifts in coding tasks. We address this issue with our proposed method, Prompt Pooling with Teacher Forcing (PP-TF), that stabilizes training by enforcing constraints on the prompt selection mechanism and leads to a 21.54{\%} improvement over Prompt Pooling. Along with the benchmark, we establish a training pipeline that can be used for CL on code models, which we believe can motivate further development of CL methods for code models.",
}
| Large-scale code generation models such as Copilot and CodeT5 have achieved impressive performance. However, libraries are upgraded or deprecated very frequently and re-training large-scale language models is computationally expensive. Therefore, Continual Learning (CL) is an important aspect that remains under-explored in the code domain. In this paper, we introduce a benchmark called CodeTask-CL that covers a wide range of tasks, including code generation, translation, summarization, and refinement, with different input and output programming languages. Next, on our CodeTask-CL benchmark, we compare popular CL techniques from NLP and Vision domains. We find that effective methods like Prompt Pooling (PP) suffer from catastrophic forgetting due to the unstable training of the prompt selection mechanism caused by stark distribution shifts in coding tasks. We address this issue with our proposed method, Prompt Pooling with Teacher Forcing (PP-TF), that stabilizes training by enforcing constraints on the prompt selection mechanism and leads to a 21.54{\%} improvement over Prompt Pooling. Along with the benchmark, we establish a training pipeline that can be used for CL on code models, which we believe can motivate further development of CL methods for code models. | [
"Yadav, Prateek",
"Sun, Qing",
"Ding, Hantian",
"Li, Xiaopeng",
"Zhang, Dejiao",
"Tan, Ming",
"Bhatia, Parminder",
"Ma, Xiaofei",
"Nallapati, Ramesh",
"Ramanathan, Murali Krishna",
"Bansal, Mohit",
"Xiang, Bing"
] | Exploring Continual Learning for Code Generation Models | acl-short.68 | Poster | 2307.02435 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.69.bib | https://aclanthology.org/2023.acl-short.69/ | @inproceedings{mirbostani-etal-2023-deep,
title = "Deep Active Learning for Morphophonological Processing",
author = "Mirbostani, Seyed Morteza and
Boreshban, Yasaman and
Khalifa, Salam and
Mirroshandel, SeyedAbolghasem and
Rambow, Owen",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.69",
doi = "10.18653/v1/2023.acl-short.69",
pages = "793--803",
abstract = "Building a system for morphological processing is a challenging task in morphologically complex languages like Arabic. Although there are some deep learning based models that achieve successful results, these models rely on a large amount of annotated data. Building such datasets, specially for some of the lower-resource Arabic dialects, is very difficult, time-consuming, and expensive. In addition, some parts of the annotated data do not contain useful information for training machine learning models. Active learning strategies allow the learner algorithm to select the most informative samples for annotation. There has been little research that focuses on applying active learning for morphological inflection and morphophonological processing. In this paper, we have proposed a deep active learning method for this task. Our experiments on Egyptian Arabic show that with only about 30{\%} of annotated data, we achieve the same results as does the state-of-the-art model on the whole dataset.",
}
| Building a system for morphological processing is a challenging task in morphologically complex languages like Arabic. Although there are some deep learning based models that achieve successful results, these models rely on a large amount of annotated data. Building such datasets, specially for some of the lower-resource Arabic dialects, is very difficult, time-consuming, and expensive. In addition, some parts of the annotated data do not contain useful information for training machine learning models. Active learning strategies allow the learner algorithm to select the most informative samples for annotation. There has been little research that focuses on applying active learning for morphological inflection and morphophonological processing. In this paper, we have proposed a deep active learning method for this task. Our experiments on Egyptian Arabic show that with only about 30{\%} of annotated data, we achieve the same results as does the state-of-the-art model on the whole dataset. | [
"Mirbostani, Seyed Morteza",
"Boreshban, Yasaman",
"Khalifa, Salam",
"Mirrosh",
"el, SeyedAbolghasem",
"Rambow, Owen"
] | Deep Active Learning for Morphophonological Processing | acl-short.69 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.70.bib | https://aclanthology.org/2023.acl-short.70/ | @inproceedings{li-etal-2023-counterfactual,
title = "Counterfactual reasoning: Testing language models{'} understanding of hypothetical scenarios",
author = "Li, Jiaxuan and
Yu, Lang and
Ettinger, Allyson",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.70",
doi = "10.18653/v1/2023.acl-short.70",
pages = "804--815",
abstract = "Current pre-trained language models have enabled remarkable improvements in downstream tasks, but it remains difficult to distinguish effects of statistical correlation from more systematic logical reasoning grounded on the understanding of real world. We tease these factors apart by leveraging counterfactual conditionals, which force language models to predict unusual consequences based on hypothetical propositions. We introduce a set of tests from psycholinguistic experiments, as well as larger-scale controlled datasets, to probe counterfactual predictions from five pre-trained language models. We find that models are consistently able to override real-world knowledge in counterfactual scenarios, and that this effect is more robust in case of stronger baseline world knowledge{---}however, we also find that for most models this effect appears largely to be driven by simple lexical cues. When we mitigate effects of both world knowledge and lexical cues to test knowledge of linguistic nuances of counterfactuals, we find that only GPT-3 shows sensitivity to these nuances, though this sensitivity is also non-trivially impacted by lexical associative factors.",
}
| Current pre-trained language models have enabled remarkable improvements in downstream tasks, but it remains difficult to distinguish effects of statistical correlation from more systematic logical reasoning grounded on the understanding of real world. We tease these factors apart by leveraging counterfactual conditionals, which force language models to predict unusual consequences based on hypothetical propositions. We introduce a set of tests from psycholinguistic experiments, as well as larger-scale controlled datasets, to probe counterfactual predictions from five pre-trained language models. We find that models are consistently able to override real-world knowledge in counterfactual scenarios, and that this effect is more robust in case of stronger baseline world knowledge{---}however, we also find that for most models this effect appears largely to be driven by simple lexical cues. When we mitigate effects of both world knowledge and lexical cues to test knowledge of linguistic nuances of counterfactuals, we find that only GPT-3 shows sensitivity to these nuances, though this sensitivity is also non-trivially impacted by lexical associative factors. | [
"Li, Jiaxuan",
"Yu, Lang",
"Ettinger, Allyson"
] | Counterfactual reasoning: Testing language models' understanding of hypothetical scenarios | acl-short.70 | Poster | 2305.16572 | [
"https://github.com/goldengua/counterfactual_inference_lm"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.71.bib | https://aclanthology.org/2023.acl-short.71/ | @inproceedings{madhani-etal-2023-bhasa,
title = "Bhasa-Abhijnaanam: Native-script and romanized Language Identification for 22 {I}ndic languages",
author = "Madhani, Yash and
Khapra, Mitesh M. and
Kunchukuttan, Anoop",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.71",
doi = "10.18653/v1/2023.acl-short.71",
pages = "816--826",
abstract = "We create publicly available language identification (LID) datasets and models in all 22 Indian languages listed in the Indian constitution in both native-script and romanized text. First, we create Bhasha-Abhijnaanam, a language identification test set for native-script as well as romanized text which spans all 22 Indic languages. We also train IndicLID, a language identifier for all the above-mentioned languages in both native and romanized script. For native-script text, it has better language coverage than existing LIDs and is competitive or better than other LIDs. IndicLID is the first LID for romanized text in Indian languages. Two major challenges for romanized text LID are the lack of training data and low-LID performance when languages are similar. We provide simple and effective solutions to these problems. In general, there has been limited work on romanized text in any language, and our findings are relevant to other languages that need romanized language identification. Our models are publicly available at \url{https://github.com/AI4Bharat/IndicLID} under open-source licenses. Our training and test sets are also publicly available at \url{https://huggingface.co/datasets/ai4bharat/Bhasha-Abhijnaanam} under open-source licenses.",
}
| We create publicly available language identification (LID) datasets and models in all 22 Indian languages listed in the Indian constitution in both native-script and romanized text. First, we create Bhasha-Abhijnaanam, a language identification test set for native-script as well as romanized text which spans all 22 Indic languages. We also train IndicLID, a language identifier for all the above-mentioned languages in both native and romanized script. For native-script text, it has better language coverage than existing LIDs and is competitive or better than other LIDs. IndicLID is the first LID for romanized text in Indian languages. Two major challenges for romanized text LID are the lack of training data and low-LID performance when languages are similar. We provide simple and effective solutions to these problems. In general, there has been limited work on romanized text in any language, and our findings are relevant to other languages that need romanized language identification. Our models are publicly available at \url{https://github.com/AI4Bharat/IndicLID} under open-source licenses. Our training and test sets are also publicly available at \url{https://huggingface.co/datasets/ai4bharat/Bhasha-Abhijnaanam} under open-source licenses. | [
"Madhani, Yash",
"Khapra, Mitesh M.",
"Kunchukuttan, Anoop"
] | Bhasa-Abhijnaanam: Native-script and romanized Language Identification for 22 Indic languages | acl-short.71 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.72.bib | https://aclanthology.org/2023.acl-short.72/ | @inproceedings{fortier-dubois-rosati-2023-using,
title = "Using contradictions improves question answering systems",
author = "Fortier-Dubois, Etienne and
Rosati, Domenic",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.72",
doi = "10.18653/v1/2023.acl-short.72",
pages = "827--840",
abstract = "This work examines the use of contradiction in natural language inference (NLI) for question answering (QA). Typically, NLI systems help answer questions by determining if a potential answer is entailed (supported) by some background context. But is it useful to also determine if an answer contradicts the context? We test this in two settings, multiple choice and extractive QA, and find that systems that incorporate contradiction can do slightly better than entailment-only systems on certain datasets. However, the best performances come from using contradiction, entailment, and QA model confidence scores together. This has implications for the deployment of QA systems in domains such as medicine and science where safety is an issue.",
}
| This work examines the use of contradiction in natural language inference (NLI) for question answering (QA). Typically, NLI systems help answer questions by determining if a potential answer is entailed (supported) by some background context. But is it useful to also determine if an answer contradicts the context? We test this in two settings, multiple choice and extractive QA, and find that systems that incorporate contradiction can do slightly better than entailment-only systems on certain datasets. However, the best performances come from using contradiction, entailment, and QA model confidence scores together. This has implications for the deployment of QA systems in domains such as medicine and science where safety is an issue. | [
"Fortier-Dubois, Etienne",
"Rosati, Domenic"
] | Using contradictions improves question answering systems | acl-short.72 | Poster | 2211.05598 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.73.bib | https://aclanthology.org/2023.acl-short.73/ | @inproceedings{peng-etal-2023-token,
title = "Token-Level Self-Evolution Training for Sequence-to-Sequence Learning",
author = "Peng, Keqin and
Ding, Liang and
Zhong, Qihuang and
Ouyang, Yuanxin and
Rong, Wenge and
Xiong, Zhang and
Tao, Dacheng",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.73",
doi = "10.18653/v1/2023.acl-short.73",
pages = "841--850",
abstract = "Adaptive training approaches, widely used in sequence-to-sequence models, commonly reweigh the losses of different target tokens based on priors, e.g. word frequency. However, most of them do not consider the variation of learning difficulty in different training steps, and overly emphasize the learning of difficult one-hot labels, making the learning deterministic and sub-optimal. In response, we present Token-Level Self-Evolution Training (SE), a simple and effective dynamic training method to fully and wisely exploit the knowledge from data. SE focuses on dynamically learning the under-explored tokens for each forward pass and adaptively regularizes the training by introducing a novel token-specific label smoothing approach. Empirically, SE yields consistent and significant improvements in three tasks, i.e. machine translation, summarization, and grammatical error correction. Encouragingly, we achieve averaging +0.93 BLEU improvement on three machine translation tasks. Analyses confirm that, besides improving lexical accuracy, SE enhances generation diversity and model generalization.",
}
| Adaptive training approaches, widely used in sequence-to-sequence models, commonly reweigh the losses of different target tokens based on priors, e.g. word frequency. However, most of them do not consider the variation of learning difficulty in different training steps, and overly emphasize the learning of difficult one-hot labels, making the learning deterministic and sub-optimal. In response, we present Token-Level Self-Evolution Training (SE), a simple and effective dynamic training method to fully and wisely exploit the knowledge from data. SE focuses on dynamically learning the under-explored tokens for each forward pass and adaptively regularizes the training by introducing a novel token-specific label smoothing approach. Empirically, SE yields consistent and significant improvements in three tasks, i.e. machine translation, summarization, and grammatical error correction. Encouragingly, we achieve averaging +0.93 BLEU improvement on three machine translation tasks. Analyses confirm that, besides improving lexical accuracy, SE enhances generation diversity and model generalization. | [
"Peng, Keqin",
"Ding, Liang",
"Zhong, Qihuang",
"Ouyang, Yuanxin",
"Rong, Wenge",
"Xiong, Zhang",
"Tao, Dacheng"
] | Token-Level Self-Evolution Training for Sequence-to-Sequence Learning | acl-short.73 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.74.bib | https://aclanthology.org/2023.acl-short.74/ | @inproceedings{yoon-etal-2023-gradient,
title = "Gradient Ascent Post-training Enhances Language Model Generalization",
author = "Yoon, Dongkeun and
Jang, Joel and
Kim, Sungdong and
Seo, Minjoon",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.74",
doi = "10.18653/v1/2023.acl-short.74",
pages = "851--864",
abstract = "In this work, we empirically show that updating pretrained LMs (350M, 1.3B, 2.7B) with just a few steps of Gradient Ascent Post-training (GAP) on random, unlabeled text corpora enhances its zero-shot generalization capabilities across diverse NLP tasks. Specifically, we show that GAP can allow LMs to become comparable to 2-3x times larger LMs across 12 different NLP tasks. We also show that applying GAP on out-of-distribution corpora leads to the most reliable performance improvements. Our findings indicate that GAP can be a promising method for improving the generalization capability of LMs without any task-specific fine-tuning.",
}
| In this work, we empirically show that updating pretrained LMs (350M, 1.3B, 2.7B) with just a few steps of Gradient Ascent Post-training (GAP) on random, unlabeled text corpora enhances its zero-shot generalization capabilities across diverse NLP tasks. Specifically, we show that GAP can allow LMs to become comparable to 2-3x times larger LMs across 12 different NLP tasks. We also show that applying GAP on out-of-distribution corpora leads to the most reliable performance improvements. Our findings indicate that GAP can be a promising method for improving the generalization capability of LMs without any task-specific fine-tuning. | [
"Yoon, Dongkeun",
"Jang, Joel",
"Kim, Sungdong",
"Seo, Minjoon"
] | Gradient Ascent Post-training Enhances Language Model Generalization | acl-short.74 | Poster | 2306.07052 | [
"https://github.com/kaist-lklab/gap"
] | https://huggingface.co/papers/2306.07052 | 2 | 0 | 0 | 4 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-short.75.bib | https://aclanthology.org/2023.acl-short.75/ | @inproceedings{burchell-etal-2023-open,
title = "An Open Dataset and Model for Language Identification",
author = "Burchell, Laurie and
Birch, Alexandra and
Bogoychev, Nikolay and
Heafield, Kenneth",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.75",
doi = "10.18653/v1/2023.acl-short.75",
pages = "865--879",
abstract = "Language identification (LID) is a fundamental step in many natural language processing pipelines. However, current LID systems are far from perfect, particularly on lower-resource languages. We present a LID model which achieves a macro-average F1 score of 0.93 and a false positive rate of 0.033{\%} across 201 languages, outperforming previous work. We achieve this by training on a curated dataset of monolingual data, which we audit manually to ensure reliability. We make both the model and the dataset available to the research community. Finally, we carry out detailed analysis into our model{'}s performance, both in comparison to existing open models and by language class.",
}
| Language identification (LID) is a fundamental step in many natural language processing pipelines. However, current LID systems are far from perfect, particularly on lower-resource languages. We present a LID model which achieves a macro-average F1 score of 0.93 and a false positive rate of 0.033{\%} across 201 languages, outperforming previous work. We achieve this by training on a curated dataset of monolingual data, which we audit manually to ensure reliability. We make both the model and the dataset available to the research community. Finally, we carry out detailed analysis into our model{'}s performance, both in comparison to existing open models and by language class. | [
"Burchell, Laurie",
"Birch, Alex",
"ra",
"Bogoychev, Nikolay",
"Heafield, Kenneth"
] | An Open Dataset and Model for Language Identification | acl-short.75 | Poster | 2305.13820 | [
"https://github.com/laurieburchell/open-lid-dataset"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.76.bib | https://aclanthology.org/2023.acl-short.76/ | @inproceedings{verma-etal-2023-evaluating,
title = "Evaluating Paraphrastic Robustness in Textual Entailment Models",
author = "Verma, Dhruv and
Lal, Yash Kumar and
Sinha, Shreyashee and
Van Durme, Benjamin and
Poliak, Adam",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.76",
doi = "10.18653/v1/2023.acl-short.76",
pages = "880--892",
abstract = "We present PaRTE, a collection of 1,126 pairs of Recognizing Textual Entailment (RTE) examples to evaluate whether models are robust to paraphrasing. We posit that if RTE models understand language, their predictions should be consistent across inputs that share the same meaning. We use the evaluation set to determine if RTE models{'} predictions change when examples are paraphrased. In our experiments, contemporary models change their predictions on 8-16{\%} of paraphrased examples, indicating that there is still room for improvement.",
}
| We present PaRTE, a collection of 1,126 pairs of Recognizing Textual Entailment (RTE) examples to evaluate whether models are robust to paraphrasing. We posit that if RTE models understand language, their predictions should be consistent across inputs that share the same meaning. We use the evaluation set to determine if RTE models{'} predictions change when examples are paraphrased. In our experiments, contemporary models change their predictions on 8-16{\%} of paraphrased examples, indicating that there is still room for improvement. | [
"Verma, Dhruv",
"Lal, Yash Kumar",
"Sinha, Shreyashee",
"Van Durme, Benjamin",
"Poliak, Adam"
] | Evaluating Paraphrastic Robustness in Textual Entailment Models | acl-short.76 | Poster | 2306.16722 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.77.bib | https://aclanthology.org/2023.acl-short.77/ | @inproceedings{tang-etal-2023-pre,
title = "Are Pre-trained Language Models Useful for Model Ensemble in {C}hinese Grammatical Error Correction?",
author = "Tang, Chenming and
Wu, Xiuyu and
Wu, Yunfang",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.77",
doi = "10.18653/v1/2023.acl-short.77",
pages = "893--901",
abstract = "Model ensemble has been in widespread use for Grammatical Error Correction (GEC), boosting model performance. We hypothesize that model ensemble based on the perplexity (PPL) computed by pre-trained language models (PLMs) should benefit the GEC system. To this end, we explore several ensemble strategies based on strong PLMs with four sophisticated single models. However, the performance does not improve but even gets worse after the PLM-based ensemble. This surprising result sets us doing a detailed analysis on the data and coming up with some insights on GEC. The human references of correct sentences is far from sufficient in the test data, and the gap between a correct sentence and an idiomatic one is worth our attention. Moreover, the PLM-based ensemble strategies provide an effective way to extend and improve GEC benchmark data. Our source code is available at \url{https://github.com/JamyDon/PLM-based-CGEC-Model-Ensemble}.",
}
| Model ensemble has been in widespread use for Grammatical Error Correction (GEC), boosting model performance. We hypothesize that model ensemble based on the perplexity (PPL) computed by pre-trained language models (PLMs) should benefit the GEC system. To this end, we explore several ensemble strategies based on strong PLMs with four sophisticated single models. However, the performance does not improve but even gets worse after the PLM-based ensemble. This surprising result sets us doing a detailed analysis on the data and coming up with some insights on GEC. The human references of correct sentences is far from sufficient in the test data, and the gap between a correct sentence and an idiomatic one is worth our attention. Moreover, the PLM-based ensemble strategies provide an effective way to extend and improve GEC benchmark data. Our source code is available at \url{https://github.com/JamyDon/PLM-based-CGEC-Model-Ensemble}. | [
"Tang, Chenming",
"Wu, Xiuyu",
"Wu, Yunfang"
] | Are Pre-trained Language Models Useful for Model Ensemble in Chinese Grammatical Error Correction? | acl-short.77 | Poster | 2305.15183 | [
"https://github.com/jamydon/plm-based-cgec-model-ensemble"
] | https://huggingface.co/papers/2305.15183 | 0 | 1 | 0 | 3 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-short.78.bib | https://aclanthology.org/2023.acl-short.78/ | @inproceedings{dixit-etal-2023-improving,
title = "Improving Factuality of Abstractive Summarization without Sacrificing Summary Quality",
author = "Dixit, Tanay and
Wang, Fei and
Chen, Muhao",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.78",
doi = "10.18653/v1/2023.acl-short.78",
pages = "902--913",
abstract = "Improving factual consistency of abstractive summarization has been a widely studied topic. However, most of the prior works on training factuality-aware models have ignored the negative effect it has on summary quality. We propose {pasted macro {`}MODEL{'}}name (i.e. Effective Factual Summarization), a candidate summary generation and ranking technique to improve summary factuality without sacrificing quality. We show that using a contrastive learning framework with our refined candidate summaries leads to significant gains on both factuality and similarity-based metrics. Specifically, we propose a ranking strategy in which we effectively combine two metrics, thereby preventing any conflict during training. Models trained using our approach show up to 6 points of absolute improvement over the base model with respect to FactCC on XSUM and 11 points on CNN/DM, without negatively affecting either similarity-based metrics or absractiveness.",
}
| Improving factual consistency of abstractive summarization has been a widely studied topic. However, most of the prior works on training factuality-aware models have ignored the negative effect it has on summary quality. We propose {pasted macro {`}MODEL{'}}name (i.e. Effective Factual Summarization), a candidate summary generation and ranking technique to improve summary factuality without sacrificing quality. We show that using a contrastive learning framework with our refined candidate summaries leads to significant gains on both factuality and similarity-based metrics. Specifically, we propose a ranking strategy in which we effectively combine two metrics, thereby preventing any conflict during training. Models trained using our approach show up to 6 points of absolute improvement over the base model with respect to FactCC on XSUM and 11 points on CNN/DM, without negatively affecting either similarity-based metrics or absractiveness. | [
"Dixit, Tanay",
"Wang, Fei",
"Chen, Muhao"
] | Improving Factuality of Abstractive Summarization without Sacrificing Summary Quality | acl-short.78 | Poster | 2305.14981 | [
"https://github.com/tanay2001/efactsum"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.79.bib | https://aclanthology.org/2023.acl-short.79/ | @inproceedings{steen-etal-2023-little,
title = "With a Little Push, {NLI} Models can Robustly and Efficiently Predict Faithfulness",
author = "Steen, Julius and
Opitz, Juri and
Frank, Anette and
Markert, Katja",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.79",
doi = "10.18653/v1/2023.acl-short.79",
pages = "914--924",
abstract = "Conditional language models still generate unfaithful output that is not supported by their input. These unfaithful generations jeopardize trust in real-world applications such as summarization or human-machine interaction, motivating a need for automatic faithfulness metrics. To implement such metrics, NLI models seem attractive, since they solve a strongly related task that comes with a wealth of prior research and data. But recent research suggests that NLI models require costly additional machinery to perform reliably across datasets, e.g., by running inference on a cartesian product of input and generated sentences, or supporting them with a question-generation/answering step. In this work we show that pure NLI models {\_}can{\_} outperform more complex metrics when combining task-adaptive data augmentation with robust inference procedures. We propose: (1) Augmenting NLI training data toadapt NL inferences to the specificities of faithfulness prediction in dialogue;(2) Making use of both entailment and contradiction probabilities in NLI, and(3) Using Monte-Carlo dropout during inference. Applied to the TRUE benchmark, which combines faithfulness datasets across diverse domains and tasks, our approach strongly improves a vanilla NLI model and significantly outperforms previous work, while showing favourable computational cost.",
}
| Conditional language models still generate unfaithful output that is not supported by their input. These unfaithful generations jeopardize trust in real-world applications such as summarization or human-machine interaction, motivating a need for automatic faithfulness metrics. To implement such metrics, NLI models seem attractive, since they solve a strongly related task that comes with a wealth of prior research and data. But recent research suggests that NLI models require costly additional machinery to perform reliably across datasets, e.g., by running inference on a cartesian product of input and generated sentences, or supporting them with a question-generation/answering step. In this work we show that pure NLI models {\_}can{\_} outperform more complex metrics when combining task-adaptive data augmentation with robust inference procedures. We propose: (1) Augmenting NLI training data toadapt NL inferences to the specificities of faithfulness prediction in dialogue;(2) Making use of both entailment and contradiction probabilities in NLI, and(3) Using Monte-Carlo dropout during inference. Applied to the TRUE benchmark, which combines faithfulness datasets across diverse domains and tasks, our approach strongly improves a vanilla NLI model and significantly outperforms previous work, while showing favourable computational cost. | [
"Steen, Julius",
"Opitz, Juri",
"Frank, Anette",
"Markert, Katja"
] | With a Little Push, NLI Models can Robustly and Efficiently Predict Faithfulness | acl-short.79 | Poster | 2305.16819 | [
"https://github.com/julmaxi/with_a_little_push"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.80.bib | https://aclanthology.org/2023.acl-short.80/ | @inproceedings{kauf-ivanova-2023-better,
title = "A Better Way to Do Masked Language Model Scoring",
author = "Kauf, Carina and
Ivanova, Anna",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.80",
doi = "10.18653/v1/2023.acl-short.80",
pages = "925--935",
abstract = "Estimating the log-likelihood of a given sentence under an autoregressive language model is straightforward: one can simply apply the chain rule and sum the log-likelihood values for each successive token. However, for masked language models (MLMs), there is no direct way to estimate the log-likelihood of a sentence. To address this issue, Salazar et al. (2020) propose to estimate sentence pseudo-log-likelihood (PLL) scores, computed by successively masking each sentence token, retrieving its score using the rest of the sentence as context, and summing the resulting values. Here, we demonstrate that the original PLL method yields inflated scores for out-of-vocabulary words and propose an adapted metric, in which we mask not only the target token, but also all within-word tokens to the right of the target. We show that our adapted metric (PLL-word-l2r) outperforms both the original PLL metric and a PLL metric in which all within-word tokens are masked. In particular, it better satisfies theoretical desiderata and better correlates with scores from autoregressive models. Finally, we show that the choice of metric affects even tightly controlled, minimal pair evaluation benchmarks (such as BLiMP), underscoring the importance of selecting an appropriate scoring metric for evaluating MLM properties.",
}
| Estimating the log-likelihood of a given sentence under an autoregressive language model is straightforward: one can simply apply the chain rule and sum the log-likelihood values for each successive token. However, for masked language models (MLMs), there is no direct way to estimate the log-likelihood of a sentence. To address this issue, Salazar et al. (2020) propose to estimate sentence pseudo-log-likelihood (PLL) scores, computed by successively masking each sentence token, retrieving its score using the rest of the sentence as context, and summing the resulting values. Here, we demonstrate that the original PLL method yields inflated scores for out-of-vocabulary words and propose an adapted metric, in which we mask not only the target token, but also all within-word tokens to the right of the target. We show that our adapted metric (PLL-word-l2r) outperforms both the original PLL metric and a PLL metric in which all within-word tokens are masked. In particular, it better satisfies theoretical desiderata and better correlates with scores from autoregressive models. Finally, we show that the choice of metric affects even tightly controlled, minimal pair evaluation benchmarks (such as BLiMP), underscoring the importance of selecting an appropriate scoring metric for evaluating MLM properties. | [
"Kauf, Carina",
"Ivanova, Anna"
] | A Better Way to Do Masked Language Model Scoring | acl-short.80 | Poster | 2305.10588 | [
"https://github.com/carina-kauf/better-mlm-scoring"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.81.bib | https://aclanthology.org/2023.acl-short.81/ | @inproceedings{heck-etal-2023-chatgpt,
title = "{C}hat{GPT} for Zero-shot Dialogue State Tracking: A Solution or an Opportunity?",
author = "Heck, Michael and
Lubis, Nurul and
Ruppik, Benjamin and
Vukovic, Renato and
Feng, Shutong and
Geishauser, Christian and
Lin, Hsien-chin and
van Niekerk, Carel and
Gasic, Milica",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.81",
doi = "10.18653/v1/2023.acl-short.81",
pages = "936--950",
abstract = "Recent research on dialog state tracking (DST) focuses on methods that allow few- and zero-shot transfer to new domains or schemas. However, performance gains heavily depend on aggressive data augmentation and fine-tuning of ever larger language model based architectures. In contrast, general purpose language models, trained on large amounts of diverse data, hold the promise of solving any kind of task without task-specific training. We present preliminary experimental results on the ChatGPT research preview, showing that ChatGPT achieves state-of-the-art performance in zero-shot DST. Despite our findings, we argue that properties inherent to general purpose models limit their ability to replace specialized systems. We further theorize that the in-context learning capabilities of such models will likely become powerful tools to support the development of dedicated dialog state trackers and enable dynamic methods.",
}
| Recent research on dialog state tracking (DST) focuses on methods that allow few- and zero-shot transfer to new domains or schemas. However, performance gains heavily depend on aggressive data augmentation and fine-tuning of ever larger language model based architectures. In contrast, general purpose language models, trained on large amounts of diverse data, hold the promise of solving any kind of task without task-specific training. We present preliminary experimental results on the ChatGPT research preview, showing that ChatGPT achieves state-of-the-art performance in zero-shot DST. Despite our findings, we argue that properties inherent to general purpose models limit their ability to replace specialized systems. We further theorize that the in-context learning capabilities of such models will likely become powerful tools to support the development of dedicated dialog state trackers and enable dynamic methods. | [
"Heck, Michael",
"Lubis, Nurul",
"Ruppik, Benjamin",
"Vukovic, Renato",
"Feng, Shutong",
"Geishauser, Christian",
"Lin, Hsien-chin",
"van Niekerk, Carel",
"Gasic, Milica"
] | ChatGPT for Zero-shot Dialogue State Tracking: A Solution or an Opportunity? | acl-short.81 | Poster | 2306.01386 | [
""
] | https://huggingface.co/papers/2306.01386 | 1 | 0 | 0 | 9 | 1 | [] | [] | [] |
https://aclanthology.org/2023.acl-short.82.bib | https://aclanthology.org/2023.acl-short.82/ | @inproceedings{chen-etal-2023-controllable,
title = "Controllable Mixed-Initiative Dialogue Generation through Prompting",
author = "Chen, Maximillian and
Yu, Xiao and
Shi, Weiyan and
Awasthi, Urvi and
Yu, Zhou",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.82",
doi = "10.18653/v1/2023.acl-short.82",
pages = "951--966",
abstract = "Mixed-initiative dialogue tasks involve repeated exchanges of information and conversational control. Conversational agents gain control by generating responses that follow particular dialogue intents or strategies, prescribed by a policy planner. The standard approach has been fine-tuning pre-trained language models to perform generation conditioned on these intents. However, these supervised generation models are limited by the cost and quality of data annotation. We instead prompt large language models as a drop-in replacement to fine-tuning on conditional generation. We formalize prompt construction for controllable mixed-initiative dialogue. Our findings show improvements over fine-tuning and ground truth responses according to human evaluation and automatic metrics for two tasks: PersuasionForGood and Emotional Support Conversations.",
}
| Mixed-initiative dialogue tasks involve repeated exchanges of information and conversational control. Conversational agents gain control by generating responses that follow particular dialogue intents or strategies, prescribed by a policy planner. The standard approach has been fine-tuning pre-trained language models to perform generation conditioned on these intents. However, these supervised generation models are limited by the cost and quality of data annotation. We instead prompt large language models as a drop-in replacement to fine-tuning on conditional generation. We formalize prompt construction for controllable mixed-initiative dialogue. Our findings show improvements over fine-tuning and ground truth responses according to human evaluation and automatic metrics for two tasks: PersuasionForGood and Emotional Support Conversations. | [
"Chen, Maximillian",
"Yu, Xiao",
"Shi, Weiyan",
"Awasthi, Urvi",
"Yu, Zhou"
] | Controllable Mixed-Initiative Dialogue Generation through Prompting | acl-short.82 | Poster | 2305.04147 | [
"https://github.com/maxlchen/controllable-mixed-initiative-dialogue-generation"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.83.bib | https://aclanthology.org/2023.acl-short.83/ | @inproceedings{mu-li-2023-enhancing,
title = "Enhancing Event Causality Identification with Counterfactual Reasoning",
author = "Mu, Feiteng and
Li, Wenjie",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.83",
doi = "10.18653/v1/2023.acl-short.83",
pages = "967--975",
abstract = "Existing methods for event causality identification (ECI) focus on mining potential causal signals, i.e., causal context keywords and event pairs. However, causal signals are ambiguous, which may lead to the context-keywords bias and the event-pairs bias. To solve this issue, we propose the \textit{counterfactual reasoning} that explicitly estimates the influence of context keywords and event pairs in training, so that we are able to eliminate the biases in inference.Experiments are conducted on two datasets, the result demonstrates the effectiveness of our method.",
}
| Existing methods for event causality identification (ECI) focus on mining potential causal signals, i.e., causal context keywords and event pairs. However, causal signals are ambiguous, which may lead to the context-keywords bias and the event-pairs bias. To solve this issue, we propose the \textit{counterfactual reasoning} that explicitly estimates the influence of context keywords and event pairs in training, so that we are able to eliminate the biases in inference.Experiments are conducted on two datasets, the result demonstrates the effectiveness of our method. | [
"Mu, Feiteng",
"Li, Wenjie"
] | Enhancing Event Causality Identification with Counterfactual Reasoning | acl-short.83 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.84.bib | https://aclanthology.org/2023.acl-short.84/ | @inproceedings{hou-etal-2023-contrastive,
title = "Contrastive Bootstrapping for Label Refinement",
author = "Hou, Shudi and
Xia, Yu and
Chen, Muhao and
Li, Sujian",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.84",
doi = "10.18653/v1/2023.acl-short.84",
pages = "976--985",
abstract = "Traditional text classification typically categorizes texts into pre-defined coarse-grained classes, from which the produced models cannot handle the real-world scenario where finer categories emerge periodically for accurate services. In this work, we investigate the setting where fine-grained classification is done only using the annotation of coarse-grained categories and the coarse-to-fine mapping. We propose a lightweight contrastive clustering-based bootstrapping method to iteratively refine the labels of passages. During clustering, it pulls away negative passage-prototype pairs under the guidance of the mapping from both global and local perspectives. Experiments on NYT and 20News show that our method outperforms the state-of-the-art methods by a large margin.",
}
| Traditional text classification typically categorizes texts into pre-defined coarse-grained classes, from which the produced models cannot handle the real-world scenario where finer categories emerge periodically for accurate services. In this work, we investigate the setting where fine-grained classification is done only using the annotation of coarse-grained categories and the coarse-to-fine mapping. We propose a lightweight contrastive clustering-based bootstrapping method to iteratively refine the labels of passages. During clustering, it pulls away negative passage-prototype pairs under the guidance of the mapping from both global and local perspectives. Experiments on NYT and 20News show that our method outperforms the state-of-the-art methods by a large margin. | [
"Hou, Shudi",
"Xia, Yu",
"Chen, Muhao",
"Li, Sujian"
] | Contrastive Bootstrapping for Label Refinement | acl-short.84 | Poster | 2306.04544 | [
"https://github.com/recorderhou/contrastive_bootstrapping_label_refinement"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.85.bib | https://aclanthology.org/2023.acl-short.85/ | @inproceedings{shode-etal-2023-nollysenti,
title = "{N}olly{S}enti: Leveraging Transfer Learning and Machine Translation for {N}igerian Movie Sentiment Classification",
author = "Shode, Iyanuoluwa and
Adelani, David Ifeoluwa and
Peng, JIng and
Feldman, Anna",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.85",
doi = "10.18653/v1/2023.acl-short.85",
pages = "986--998",
abstract = "Africa has over 2000 indigenous languages but they are under-represented in NLP research due to lack of datasets. In recent years, there have been progress in developing labelled corpora for African languages. However, they are often available in a single domain and may not generalize to other domains. In this paper, we focus on the task of sentiment classification for cross-domain adaptation. We create a new dataset, Nollywood movie reviews for five languages widely spoken in Nigeria (English, Hausa, Igbo, Nigerian Pidgin, and Yoruba). We provide an extensive empirical evaluation using classical machine learning methods and pre-trained language models. By leveraging transfer learning, we compare the performance of cross-domain adaptation from Twitter domain, and cross-lingual adaptation from English language. Our evaluation shows that transfer from English in the same target domain leads to more than 5{\%} improvement in accuracy compared to transfer from Twitter in the same language. To further mitigate the domain difference, we leverage machine translation from English to other Nigerian languages, which leads to a further improvement of 7{\%} over cross-lingual evaluation. While machine translation to low-resource languages are often of low quality, our analysis shows that sentiment related words are often preserved.",
}
| Africa has over 2000 indigenous languages but they are under-represented in NLP research due to lack of datasets. In recent years, there have been progress in developing labelled corpora for African languages. However, they are often available in a single domain and may not generalize to other domains. In this paper, we focus on the task of sentiment classification for cross-domain adaptation. We create a new dataset, Nollywood movie reviews for five languages widely spoken in Nigeria (English, Hausa, Igbo, Nigerian Pidgin, and Yoruba). We provide an extensive empirical evaluation using classical machine learning methods and pre-trained language models. By leveraging transfer learning, we compare the performance of cross-domain adaptation from Twitter domain, and cross-lingual adaptation from English language. Our evaluation shows that transfer from English in the same target domain leads to more than 5{\%} improvement in accuracy compared to transfer from Twitter in the same language. To further mitigate the domain difference, we leverage machine translation from English to other Nigerian languages, which leads to a further improvement of 7{\%} over cross-lingual evaluation. While machine translation to low-resource languages are often of low quality, our analysis shows that sentiment related words are often preserved. | [
"Shode, Iyanuoluwa",
"Adelani, David Ifeoluwa",
"Peng, JIng",
"Feldman, Anna"
] | NollySenti: Leveraging Transfer Learning and Machine Translation for Nigerian Movie Sentiment Classification | acl-short.85 | Oral | 2305.10971 | [
"https://github.com/iyanush/nollysenti"
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.86.bib | https://aclanthology.org/2023.acl-short.86/ | @inproceedings{mensah-etal-2023-trading,
title = "Trading Syntax Trees for Wordpieces: Target-oriented Opinion Words Extraction with Wordpieces and Aspect Enhancement",
author = "Mensah, Samuel and
Sun, Kai and
Aletras, Nikolaos",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.86",
doi = "10.18653/v1/2023.acl-short.86",
pages = "999--1007",
abstract = "State-of-the-art target-oriented opinion word extraction (TOWE) models typically use BERT-based text encoders that operate on the word level, along with graph convolutional networks (GCNs) that incorporate syntactic information extracted from syntax trees. These methods achieve limited gains with GCNs and have difficulty using BERT wordpieces. Meanwhile, BERT wordpieces are known to be effective at representing rare words or words with insufficient context information. To address this issue, this work trades syntax trees for BERT wordpieces by entirely removing the GCN component from the methods{'} architectures. To enhance TOWE performance, we tackle the issue of aspect representation loss during encoding. Instead of solely utilizing a sentence as the input, we use a sentence-aspect pair. Our relatively simple approach achieves state-of-the-art results on benchmark datasets and should serve as a strong baseline for further research.",
}
| State-of-the-art target-oriented opinion word extraction (TOWE) models typically use BERT-based text encoders that operate on the word level, along with graph convolutional networks (GCNs) that incorporate syntactic information extracted from syntax trees. These methods achieve limited gains with GCNs and have difficulty using BERT wordpieces. Meanwhile, BERT wordpieces are known to be effective at representing rare words or words with insufficient context information. To address this issue, this work trades syntax trees for BERT wordpieces by entirely removing the GCN component from the methods{'} architectures. To enhance TOWE performance, we tackle the issue of aspect representation loss during encoding. Instead of solely utilizing a sentence as the input, we use a sentence-aspect pair. Our relatively simple approach achieves state-of-the-art results on benchmark datasets and should serve as a strong baseline for further research. | [
"Mensah, Samuel",
"Sun, Kai",
"Aletras, Nikolaos"
] | Trading Syntax Trees for Wordpieces: Target-oriented Opinion Words Extraction with Wordpieces and Aspect Enhancement | acl-short.86 | Poster | 2305.11034 | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
|
https://aclanthology.org/2023.acl-short.87.bib | https://aclanthology.org/2023.acl-short.87/ | @inproceedings{jimerson-etal-2023-unhelpful,
title = "An (unhelpful) guide to selecting the best {ASR} architecture for your under-resourced language",
author = "Jimerson, Robert and
Liu, Zoey and
Prud{'}hommeaux, Emily",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.87",
doi = "10.18653/v1/2023.acl-short.87",
pages = "1008--1016",
abstract = "Advances in deep neural models for automatic speech recognition (ASR) have yielded dramatic improvements in ASR quality for resource-rich languages, with English ASR now achieving word error rates comparable to that of human transcribers. The vast majority of the world{'}s languages, however, lack the quantity of data necessary to approach this level of accuracy. In this paper we use four of the most popular ASR toolkits to train ASR models for eleven languages with limited ASR training resources: eleven widely spoken languages of Africa, Asia, and South America, one endangered language of Central America, and three critically endangered languages of North America. We find that no single architecture consistently outperforms any other. These differences in performance so far do not appear to be related to any particular feature of the datasets or characteristics of the languages. These findings have important implications for future research in ASR for under-resourced languages. ASR systems for languages with abundant existing media and available speakers may derive the most benefit simply by collecting large amounts of additional acoustic and textual training data. Communities using ASR to support endangered language documentation efforts, who cannot easily collect more data, might instead focus on exploring multiple architectures and hyperparameterizations to optimize performance within the constraints of their available data and resources.",
}
| Advances in deep neural models for automatic speech recognition (ASR) have yielded dramatic improvements in ASR quality for resource-rich languages, with English ASR now achieving word error rates comparable to that of human transcribers. The vast majority of the world{'}s languages, however, lack the quantity of data necessary to approach this level of accuracy. In this paper we use four of the most popular ASR toolkits to train ASR models for eleven languages with limited ASR training resources: eleven widely spoken languages of Africa, Asia, and South America, one endangered language of Central America, and three critically endangered languages of North America. We find that no single architecture consistently outperforms any other. These differences in performance so far do not appear to be related to any particular feature of the datasets or characteristics of the languages. These findings have important implications for future research in ASR for under-resourced languages. ASR systems for languages with abundant existing media and available speakers may derive the most benefit simply by collecting large amounts of additional acoustic and textual training data. Communities using ASR to support endangered language documentation efforts, who cannot easily collect more data, might instead focus on exploring multiple architectures and hyperparameterizations to optimize performance within the constraints of their available data and resources. | [
"Jimerson, Robert",
"Liu, Zoey",
"Prud{'}hommeaux, Emily"
] | An (unhelpful) guide to selecting the best ASR architecture for your under-resourced language | acl-short.87 | Oral | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.88.bib | https://aclanthology.org/2023.acl-short.88/ | @inproceedings{orlikowski-etal-2023-ecological,
title = "The Ecological Fallacy in Annotation: Modeling Human Label Variation goes beyond Sociodemographics",
author = {Orlikowski, Matthias and
R{\"o}ttger, Paul and
Cimiano, Philipp and
Hovy, Dirk},
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.88",
doi = "10.18653/v1/2023.acl-short.88",
pages = "1017--1029",
abstract = "Many NLP tasks exhibit human label variation, where different annotators give different labels to the same texts. This variation is known to depend, at least in part, on the sociodemographics of annotators. Recent research aims to model individual annotator behaviour rather than predicting aggregated labels, and we would expect that sociodemographic information is useful for these models. On the other hand, the ecological fallacy states that aggregate group behaviour, such as the behaviour of the average female annotator, does not necessarily explain individual behaviour. To account for sociodemographics in models of individual annotator behaviour, we introduce group-specific layers to multi-annotator models. In a series of experiments for toxic content detection, we find that explicitly accounting for sociodemographic attributes in this way does not significantly improve model performance. This result shows that individual annotation behaviour depends on much more than just sociodemographics.",
}
| Many NLP tasks exhibit human label variation, where different annotators give different labels to the same texts. This variation is known to depend, at least in part, on the sociodemographics of annotators. Recent research aims to model individual annotator behaviour rather than predicting aggregated labels, and we would expect that sociodemographic information is useful for these models. On the other hand, the ecological fallacy states that aggregate group behaviour, such as the behaviour of the average female annotator, does not necessarily explain individual behaviour. To account for sociodemographics in models of individual annotator behaviour, we introduce group-specific layers to multi-annotator models. In a series of experiments for toxic content detection, we find that explicitly accounting for sociodemographic attributes in this way does not significantly improve model performance. This result shows that individual annotation behaviour depends on much more than just sociodemographics. | [
"Orlikowski, Matthias",
"R{\\\"o}ttger, Paul",
"Cimiano, Philipp",
"Hovy, Dirk"
] | The Ecological Fallacy in Annotation: Modeling Human Label Variation goes beyond Sociodemographics | acl-short.88 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
||
https://aclanthology.org/2023.acl-short.89.bib | https://aclanthology.org/2023.acl-short.89/ | @inproceedings{bhargava-penn-2023-decomposed,
title = "Decomposed scoring of {CCG} dependencies",
author = "Bhargava, Aditya and
Penn, Gerald",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-short.89",
doi = "10.18653/v1/2023.acl-short.89",
pages = "1030--1040",
abstract = "In statistical parsing with CCG, the standard evaluation method is based on predicate-argument structure and evaluates dependencies labelled in part by lexical categories. When a predicate has multiple argument slots that can be filled, the same lexical category is used for the label of multiple dependencies. In this paper, we show that this evaluation can result in disproportionate penalization of supertagging errors and obfuscate the truly erroneous dependencies. Enabled by the compositional nature of CCG lexical categories, we propose *decomposed scoring* based on subcategorial labels to address this. To evaluate our scoring method, we engage fellow categorial grammar researchers in two English-language judgement tasks: (1) directly ranking the outputs of the standard and experimental scoring methods; and (2) determining which of two sentences has the better parse in cases where the two scoring methods disagree on their ranks. Overall, the judges prefer decomposed scoring in each task; but there is substantial disagreement among the judges in 24{\%} of the given cases, pointing to potential issues with parser evaluations in general.",
}
| In statistical parsing with CCG, the standard evaluation method is based on predicate-argument structure and evaluates dependencies labelled in part by lexical categories. When a predicate has multiple argument slots that can be filled, the same lexical category is used for the label of multiple dependencies. In this paper, we show that this evaluation can result in disproportionate penalization of supertagging errors and obfuscate the truly erroneous dependencies. Enabled by the compositional nature of CCG lexical categories, we propose *decomposed scoring* based on subcategorial labels to address this. To evaluate our scoring method, we engage fellow categorial grammar researchers in two English-language judgement tasks: (1) directly ranking the outputs of the standard and experimental scoring methods; and (2) determining which of two sentences has the better parse in cases where the two scoring methods disagree on their ranks. Overall, the judges prefer decomposed scoring in each task; but there is substantial disagreement among the judges in 24{\%} of the given cases, pointing to potential issues with parser evaluations in general. | [
"Bhargava, Aditya",
"Penn, Gerald"
] | Decomposed scoring of CCG dependencies | acl-short.89 | Poster | [
""
] | -1 | -1 | -1 | -1 | 0 | [] | [] | [] |
Subsets and Splits