Titles
stringlengths 6
220
| Abstracts
stringlengths 37
3.26k
| Years
int64 1.99k
2.02k
| Categories
stringclasses 1
value |
|---|---|---|---|
Extending Neural Keyword Extraction with TF-IDF tagset matching
|
Keyword extraction is the task of identifying words (or multi-word
expressions) that best describe a given document and serve in news portals to
link articles of similar topics. In this work we develop and evaluate our
methods on four novel data sets covering less represented, morphologically-rich
languages in European news media industry (Croatian, Estonian, Latvian and
Russian). First, we perform evaluation of two supervised neural
transformer-based methods (TNT-KID and BERT+BiLSTM CRF) and compare them to a
baseline TF-IDF based unsupervised approach. Next, we show that by combining
the keywords retrieved by both neural transformer based methods and extending
the final set of keywords with an unsupervised TF-IDF based technique, we can
drastically improve the recall of the system, making it appropriate to be used
as a recommendation system in the media house environment.
| 2,022 |
Computation and Language
|
Decoupling the Role of Data, Attention, and Losses in Multimodal
Transformers
|
Recently multimodal transformer models have gained popularity because their
performance on language and vision tasks suggest they learn rich
visual-linguistic representations. Focusing on zero-shot image retrieval tasks,
we study three important factors which can impact the quality of learned
representations: pretraining data, the attention mechanism, and loss functions.
By pretraining models on six datasets, we observe that dataset noise and
language similarity to our downstream task are important indicators of model
performance. Through architectural analysis, we learn that models with a
multimodal attention mechanism can outperform deeper models with modality
specific attention mechanisms. Finally, we show that successful contrastive
losses used in the self-supervised learning literature do not yield similar
performance gains when used in multimodal transformers
| 2,021 |
Computation and Language
|
Short Text Clustering with Transformers
|
Recent techniques for the task of short text clustering often rely on word
embeddings as a transfer learning component. This paper shows that sentence
vector representations from Transformers in conjunction with different
clustering methods can be successfully applied to address the task.
Furthermore, we demonstrate that the algorithm of enhancement of clustering via
iterative classification can further improve initial clustering performance
with different classifiers, including those based on pre-trained Transformer
language models.
| 2,021 |
Computation and Language
|
Neural OCR Post-Hoc Correction of Historical Corpora
|
Optical character recognition (OCR) is crucial for a deeper access to
historical collections. OCR needs to account for orthographic variations,
typefaces, or language evolution (i.e., new letters, word spellings), as the
main source of character, word, or word segmentation transcription errors. For
digital corpora of historical prints, the errors are further exacerbated due to
low scan quality and lack of language standardization.
For the task of OCR post-hoc correction, we propose a neural approach based
on a combination of recurrent (RNN) and deep convolutional network (ConvNet) to
correct OCR transcription errors. At character level we flexibly capture
errors, and decode the corrected output based on a novel attention mechanism.
Accounting for the input and output similarity, we propose a new loss function
that rewards the model's correcting behavior.
Evaluation on a historical book corpus in German language shows that our
models are robust in capturing diverse OCR transcription errors and reduce the
word error rate of 32.3% by more than 89%.
| 2,021 |
Computation and Language
|
The Harrington Yowlumne Narrative Corpus
|
Minority languages continue to lack adequate resources for their development,
especially in the technological domain. Likewise, the J.P. Harrington Papers
collection at the Smithsonian Institution are difficult to access in practical
terms for community members and researchers due to its handwritten and
disorganized format. Our current work seeks to make a portion of this
publicly-available yet problematic material practically accessible for natural
language processing use. Here, we present the Harrington Yowlumne Narrative
Corpus, a corpus of 20 narrative texts that derive from the Tejone\~no Yowlumne
community of the Tinliw rancheria in Kern County, California between 1910 and
1925. We digitally transcribe the texts and, through a Levenshtein
distance-based algorithm and manual checking, we provide gold-standard aligned
normalized and lemmatized text. We likewise provide POS tags for each
lemmatized token via a lexicon-based deterministic approach. Altogether, the
corpus contains 57,136 transcribed characters aligned with 10,719 gold standard
text-normalized words.
| 2,022 |
Computation and Language
|
Polyphone Disambiguition in Mandarin Chinese with Semi-Supervised
Learning
|
The majority of Chinese characters are monophonic, while a special group of
characters, called polyphonic characters, have multiple pronunciations. As a
prerequisite of performing speech-related generative tasks, the correct
pronunciation must be identified among several candidates. This process is
called Polyphone Disambiguation. Although the problem has been well explored
with both knowledge-based and learning-based approaches, it remains challenging
due to the lack of publicly available labeled datasets and the irregular nature
of polyphone in Mandarin Chinese. In this paper, we propose a novel
semi-supervised learning (SSL) framework for Mandarin Chinese polyphone
disambiguation that can potentially leverage unlimited unlabeled text data. We
explore the effect of various proxy labeling strategies including
entropy-thresholding and lexicon-based labeling. Qualitative and quantitative
experiments demonstrate that our method achieves state-of-the-art performance.
In addition, we publish a novel dataset specifically for the polyphone
disambiguation task to promote further researches.
| 2,021 |
Computation and Language
|
Commonsense Knowledge Mining from Term Definitions
|
Commonsense knowledge has proven to be beneficial to a variety of application
areas, including question answering and natural language understanding.
Previous work explored collecting commonsense knowledge triples automatically
from text to increase the coverage of current commonsense knowledge graphs. We
investigate a few machine learning approaches to mining commonsense knowledge
triples using dictionary term definitions as inputs and provide some initial
evaluation of the results. We start from extracting candidate triples using
part-of-speech tag patterns from text, and then compare the performance of
three existing models for triple scoring. Our experiments show that term
definitions contain some valid and novel commonsense knowledge triples for some
semantic relations, and also indicate some challenges with using existing
triple scoring models.
| 2,021 |
Computation and Language
|
Hierarchical Ranking for Answer Selection
|
Answer selection is a task to choose the positive answers from a pool of
candidate answers for a given question. In this paper, we propose a novel
strategy for answer selection, called hierarchical ranking. We introduce three
levels of ranking: point-level ranking, pair-level ranking, and list-level
ranking. They formulate their optimization objectives by employing supervisory
information from different perspectives to achieve the same goal of ranking
candidate answers. Therefore, the three levels of ranking are related and they
can promote each other. We take the well-performed compare-aggregate model as
the backbone and explore three schemes to implement the idea of applying the
hierarchical rankings jointly: the scheme under the Multi-Task Learning (MTL)
strategy, the Ranking Integration (RI) scheme, and the Progressive Ranking
Integration (PRI) scheme. Experimental results on two public datasets, WikiQA
and TREC-QA, demonstrate that the proposed hierarchical ranking is effective.
Our method achieves state-of-the-art (non-BERT) performance on both TREC-QA and
WikiQA.
| 2,021 |
Computation and Language
|
GTAE: Graph-Transformer based Auto-Encoders for Linguistic-Constrained
Text Style Transfer
|
Non-parallel text style transfer has attracted increasing research interests
in recent years. Despite successes in transferring the style based on the
encoder-decoder framework, current approaches still lack the ability to
preserve the content and even logic of original sentences, mainly due to the
large unconstrained model space or too simplified assumptions on latent
embedding space. Since language itself is an intelligent product of humans with
certain grammars and has a limited rule-based model space by its nature,
relieving this problem requires reconciling the model capacity of deep neural
networks with the intrinsic model constraints from human linguistic rules. To
this end, we propose a method called Graph Transformer based Auto Encoder
(GTAE), which models a sentence as a linguistic graph and performs feature
extraction and style transfer at the graph level, to maximally retain the
content and the linguistic structure of original sentences. Quantitative
experiment results on three non-parallel text style transfer tasks show that
our model outperforms state-of-the-art methods in content preservation, while
achieving comparable performance on transfer accuracy and sentence naturalness.
| 2,021 |
Computation and Language
|
Many Hands Make Light Work: Using Essay Traits to Automatically Score
Essays
|
Most research in the area of automatic essay grading (AEG) is geared towards
scoring the essay holistically while there has also been some work done on
scoring individual essay traits. In this paper, we describe a way to score
essays holistically using a multi-task learning (MTL) approach, where scoring
the essay holistically is the primary task, and scoring the essay traits is the
auxiliary task. We compare our results with a single-task learning (STL)
approach, using both LSTMs and BiLSTMs. We also compare our results of the
auxiliary task with such tasks done in other AEG systems. To find out which
traits work best for different types of essays, we conduct ablation tests for
each of the essay traits. We also report the runtime and number of training
parameters for each system. We find that MTL-based BiLSTM system gives the best
results for scoring the essay holistically, as well as performing well on
scoring the essay traits.
| 2,021 |
Computation and Language
|
VRoC: Variational Autoencoder-aided Multi-task Rumor Classifier Based on
Text
|
Social media became popular and percolated almost all aspects of our daily
lives. While online posting proves very convenient for individual users, it
also fosters fast-spreading of various rumors. The rapid and wide percolation
of rumors can cause persistent adverse or detrimental impacts. Therefore,
researchers invest great efforts on reducing the negative impacts of rumors.
Towards this end, the rumor classification system aims to detect, track, and
verify rumors in social media. Such systems typically include four components:
(i) a rumor detector, (ii) a rumor tracker, (iii) a stance classifier, and (iv)
a veracity classifier. In order to improve the state-of-the-art in rumor
detection, tracking, and verification, we propose VRoC, a tweet-level
variational autoencoder-based rumor classification system. VRoC consists of a
co-train engine that trains variational autoencoders (VAEs) and rumor
classification components. The co-train engine helps the VAEs to tune their
latent representations to be classifier-friendly. We also show that VRoC is
able to classify unseen rumors with high levels of accuracy. For the PHEME
dataset, VRoC consistently outperforms several state-of-the-art techniques, on
both observed and unobserved rumors, by up to 26.9%, in terms of macro-F1
scores.
| 2,021 |
Computation and Language
|
Metric-Type Identification for Multi-Level Header Numerical Tables in
Scientific Papers
|
Numerical tables are widely used to present experimental results in
scientific papers. For table understanding, a metric-type is essential to
discriminate numbers in the tables. We introduce a new information extraction
task, metric-type identification from multi-level header numerical tables, and
provide a dataset extracted from scientific papers consisting of header tables,
captions, and metric-types. We then propose two joint-learning neural
classification and generation schemes featuring pointer-generator-based and
BERT-based models. Our results show that the joint models can handle both
in-header and out-of-header metric-type identification problems.
| 2,021 |
Computation and Language
|
Fine-tuning BERT-based models for Plant Health Bulletin Classification
|
In the era of digitization, different actors in agriculture produce numerous
data. Such data contains already latent historical knowledge in the domain.
This knowledge enables us to precisely study natural hazards within global or
local aspects, and then improve the risk prevention tasks and augment the
yield, which helps to tackle the challenge of growing population and changing
alimentary habits. In particular, French Plants Health Bulletins (BSV, for its
name in French Bulletin de Sant{\'e} du V{\'e}g{\'e}tal) give information about
the development stages of phytosanitary risks in agricultural production.
However, they are written in natural language, thus, machines and human cannot
exploit them as efficiently as it could be. Natural language processing (NLP)
technologies aim to automatically process and analyze large amounts of natural
language data. Since the 2010s, with the increases in computational power and
parallelization, representation learning and deep learning methods became
widespread in NLP. Recent advancements Bidirectional Encoder Representations
from Transformers (BERT) inspire us to rethink of knowledge representation and
natural language understanding in plant health management domain. The goal in
this work is to propose a BERT-based approach to automatically classify the BSV
to make their data easily indexable. We sampled 200 BSV to finetune the
pretrained BERT language models and classify them as pest or/and disease and we
show preliminary results.
| 2,021 |
Computation and Language
|
LSTM-SAKT: LSTM-Encoded SAKT-like Transformer for Knowledge Tracing
|
This paper introduces the 2nd place solution for the Riiid! Answer
Correctness Prediction in Kaggle, the world's largest data science competition
website. This competition was held from October 16, 2020, to January 7, 2021,
with 3395 teams and 4387 competitors. The main insights and contributions of
this paper are as follows. (i) We pointed out existing Transformer-based models
are suffering from a problem that the information which their query/key/value
can contain is limited. To solve this problem, we proposed a method that uses
LSTM to obtain query/key/value and verified its effectiveness. (ii) We pointed
out 'inter-container' leakage problem, which happens in datasets where
questions are sometimes served together. To solve this problem, we showed
special indexing/masking techniques that are useful when using RNN-variants and
Transformer. (iii) We found additional hand-crafted features are effective to
overcome the limits of Transformer, which can never consider the samples older
than the sequence length.
| 2,021 |
Computation and Language
|
Gamified Crowdsourcing for Idiom Corpora Construction
|
Learning idiomatic expressions is seen as one of the most challenging stages
in second language learning because of their unpredictable meaning. A similar
situation holds for their identification within natural language processing
applications such as machine translation and parsing. The lack of high-quality
usage samples exacerbates this challenge not only for humans but also for
artificial intelligence systems. This article introduces a gamified
crowdsourcing approach for collecting language learning materials for idiomatic
expressions; a messaging bot is designed as an asynchronous multiplayer game
for native speakers who compete with each other while providing idiomatic and
nonidiomatic usage examples and rating other players' entries. As opposed to
classical crowdprocessing annotation efforts in the field, for the first time
in the literature, a crowdcreating & crowdrating approach is implemented and
tested for idiom corpora construction. The approach is language independent and
evaluated on two languages in comparison to traditional data preparation
techniques in the field. The reaction of the crowd is monitored under different
motivational means (namely, gamification affordances and monetary rewards). The
results reveal that the proposed approach is powerful in collecting the
targeted materials, and although being an explicit crowdsourcing approach, it
is found entertaining and useful by the crowd. The approach has been shown to
have the potential to speed up the construction of idiom corpora for different
natural languages to be used as second language learning material, training
data for supervised idiom identification systems, or samples for lexicographic
studies.
| 2,022 |
Computation and Language
|
Multilingual LAMA: Investigating Knowledge in Multilingual Pretrained
Language Models
|
Recently, it has been found that monolingual English language models can be
used as knowledge bases. Instead of structural knowledge base queries, masked
sentences such as "Paris is the capital of [MASK]" are used as probes. We
translate the established benchmarks TREx and GoogleRE into 53 languages.
Working with mBERT, we investigate three questions. (i) Can mBERT be used as a
multilingual knowledge base? Most prior work only considers English. Extending
research to multiple languages is important for diversity and accessibility.
(ii) Is mBERT's performance as knowledge base language-independent or does it
vary from language to language? (iii) A multilingual model is trained on more
text, e.g., mBERT is trained on 104 Wikipedias. Can mBERT leverage this for
better performance? We find that using mBERT as a knowledge base yields varying
performance across languages and pooling predictions across languages improves
performance. Conversely, mBERT exhibits a language bias; e.g., when queried in
Italian, it tends to predict Italy as the country of origin.
| 2,021 |
Computation and Language
|
Text-to-hashtag Generation using Seq2seq Learning
|
In this paper, we studied whether models based on BiLSTM and BERT can predict
hashtags in Brazilian Portuguese for Ecommerce websites. Hashtags have a
sizable financial impact on Ecommerce. We processed a corpus of Ecommerce
reviews as inputs, and predicted hashtags as outputs. We evaluated the results
using four quantitative metrics: NIST, BLEU, METEOR and a crowdsourced score. A
word cloud was used as a qualitative metric. While all computer-generated
metrics (NIST, BLEU and METEOR) indicated bad results, the crowdsourced results
produced amazing scores. We concluded that the texts predicted by the neural
networks are very promising for use as hashtags for products on Ecommerce
websites. The code for this work is available at
https://github.com/augustocamargo/text-to-hashtag.
| 2,021 |
Computation and Language
|
Transfer Learning Approach for Detecting Psychological Distress in
Brexit Tweets
|
In 2016, United Kingdom (UK) citizens voted to leave the European Union (EU),
which was officially implemented in 2020. During this period, UK residents
experienced a great deal of uncertainty around the UK's continued relationship
with the EU. Many people have used social media platforms to express their
emotions about this critical event. Sentiment analysis has been recently
considered as an important tool for detecting mental well-being in Twitter
contents. However, detecting the psychological distress status in
political-related tweets is a challenging task due to the lack of explicit
sentences describing the depressive or anxiety status. To address this problem,
this paper leverages a transfer learning approach for sentiment analysis to
measure the non-clinical psychological distress status in Brexit tweets. The
framework transfers the knowledge learnt from self-reported psychological
distress tweets (source domain) to detect the distress status in Brexit tweets
(target domain). The framework applies a domain adaptation technique to
decrease the impact of negative transfer between source and target domains. The
paper also introduces a Brexit distress index that can be used to detect levels
of psychological distress of individuals in Brexit tweets. We design an
experiment that includes data from both domains. The proposed model is able to
detect the non-clinical psychological distress status in Brexit tweets with an
accuracy of 66% and 62% on the source and target domains, respectively.
| 2,021 |
Computation and Language
|
Counting Protests in News Articles: A Dataset and Semi-Automated Data
Collection Pipeline
|
Between January 2017 and January 2021, thousands of local news sources in the
United States reported on over 42,000 protests about topics such as civil
rights, immigration, guns, and the environment. Given the vast number of local
journalists that report on protests daily, extracting these events as
structured data to understand temporal and geographic trends can empower civic
decision-making. However, the task of extracting events from news articles
presents well known challenges to the NLP community in the fields of domain
detection, slot filling, and coreference resolution.
To help improve the resources available for extracting structured data from
news stories, our contribution is three-fold. We 1) release a manually labeled
dataset of news article URLs, dates, locations, crowd size estimates, and 494
discrete descriptive tags corresponding to 42,347 reported protest events in
the United States between January 2017 and January 2021; 2) describe the
semi-automated data collection pipeline used to discover, sort, and review the
144,568 English articles that comprise the dataset; and 3) benchmark a
long-short term memory (LSTM) low dimensional classifier that demonstrates the
utility of processing news articles based on syntactic structures, such as
paragraphs and sentences, to count the number of reported protest events.
| 2,021 |
Computation and Language
|
Revisiting the Prepositional-Phrase Attachment Problem Using Explicit
Commonsense Knowledge
|
We revisit the challenging problem of resolving prepositional-phrase (PP)
attachment ambiguity. To date, proposed solutions are either rule-based, where
explicit grammar rules direct how to resolve ambiguities; or statistical, where
the decision is learned from a corpus of labeled examples. We argue that
explicit commonsense knowledge bases can provide an essential ingredient for
making good attachment decisions. We implemented a module, named Patch-Comm,
that can be used by a variety of conventional parsers, to make attachment
decisions. Where the commonsense KB does not provide direct answers, we fall
back on a more general system that infers "out-of-knowledge-base" assertions in
a manner similar to the way some NLP systems handle out-of-vocabulary words.
Our results suggest that the commonsense knowledge-based approach can provide
the best of both worlds, integrating rule-based and statistical techniques. As
the field is increasingly coming to recognize the importance of explainability
in AI, a commonsense approach can enable NLP developers to better understand
the behavior of systems, and facilitate natural dialogues with end users.
| 2,021 |
Computation and Language
|
End2End Acoustic to Semantic Transduction
|
In this paper, we propose a novel end-to-end sequence-to-sequence spoken
language understanding model using an attention mechanism. It reliably selects
contextual acoustic features in order to hypothesize semantic contents. An
initial architecture capable of extracting all pronounced words and concepts
from acoustic spans is designed and tested. With a shallow fusion language
model, this system reaches a 13.6 concept error rate (CER) and an 18.5 concept
value error rate (CVER) on the French MEDIA corpus, achieving an absolute 2.8
points reduction compared to the state-of-the-art. Then, an original model is
proposed for hypothesizing concepts and their values. This transduction reaches
a 15.4 CER and a 21.6 CVER without any new type of context.
| 2,021 |
Computation and Language
|
Measuring and Improving Consistency in Pretrained Language Models
|
Consistency of a model -- that is, the invariance of its behavior under
meaning-preserving alternations in its input -- is a highly desirable property
in natural language processing. In this paper we study the question: Are
Pretrained Language Models (PLMs) consistent with respect to factual knowledge?
To this end, we create ParaRel, a high-quality resource of cloze-style query
English paraphrases. It contains a total of 328 paraphrases for 38 relations.
Using ParaRel, we show that the consistency of all PLMs we experiment with is
poor -- though with high variance between relations. Our analysis of the
representational spaces of PLMs suggests that they have a poor structure and
are currently not suitable for representing knowledge robustly. Finally, we
propose a method for improving model consistency and experimentally demonstrate
its effectiveness.
| 2,021 |
Computation and Language
|
SJ_AJ@DravidianLangTech-EACL2021: Task-Adaptive Pre-Training of
Multilingual BERT models for Offensive Language Identification
|
In this paper we present our submission for the EACL 2021-Shared Task on
Offensive Language Identification in Dravidian languages. Our final system is
an ensemble of mBERT and XLM-RoBERTa models which leverage task-adaptive
pre-training of multilingual BERT models with a masked language modeling
objective. Our system was ranked 1st for Kannada, 2nd for Malayalam and 3rd for
Tamil.
| 2,021 |
Computation and Language
|
Do Question Answering Modeling Improvements Hold Across Benchmarks?
|
Do question answering (QA) modeling improvements (e.g., choice of
architecture and training procedure) hold consistently across the diverse
landscape of QA benchmarks? To study this question, we introduce the notion of
concurrence -- two benchmarks have high concurrence on a set of modeling
approaches if they rank the modeling approaches similarly. We measure the
concurrence between 32 QA benchmarks on a set of 20 diverse modeling approaches
and find that human-constructed benchmarks have high concurrence amongst
themselves, even if their passage and question distributions are very
different. Surprisingly, even downsampled human-constructed benchmarks (i.e.,
collecting less data) and programmatically-generated benchmarks (e.g.,
cloze-formatted examples) have high concurrence with human-constructed
benchmarks. These results indicate that, despite years of intense community
focus on a small number of benchmarks, the modeling improvements studied hold
broadly.
| 2,023 |
Computation and Language
|
Improving Distantly-Supervised Relation Extraction through BERT-based
Label & Instance Embeddings
|
Distantly-supervised relation extraction (RE) is an effective method to scale
RE to large corpora but suffers from noisy labels. Existing approaches try to
alleviate noise through multi-instance learning and by providing additional
information, but manage to recognize mainly the top frequent relations,
neglecting those in the long-tail. We propose REDSandT (Relation Extraction
with Distant Supervision and Transformers), a novel distantly-supervised
transformer-based RE method, that manages to capture a wider set of relations
through highly informative instance and label embeddings for RE, by exploiting
BERT's pre-trained model, and the relationship between labels and entities,
respectively. We guide REDSandT to focus solely on relational tokens by
fine-tuning BERT on a structured input, including the sub-tree connecting an
entity pair and the entities' types. Using the extracted informative vectors,
we shape label embeddings, which we also use as attention mechanism over
instances to further reduce noise. Finally, we represent sentences by
concatenating relation and instance embeddings. Experiments in the NYT-10
dataset show that REDSandT captures a broader set of relations with higher
confidence, achieving state-of-the-art AUC (0.424).
| 2,021 |
Computation and Language
|
Generative Spoken Language Modeling from Raw Audio
|
We introduce Generative Spoken Language Modeling, the task of learning the
acoustic and linguistic characteristics of a language from raw audio (no text,
no labels), and a set of metrics to automatically evaluate the learned
representations at acoustic and linguistic levels for both encoding and
generation. We set up baseline systems consisting of a discrete speech encoder
(returning pseudo-text units), a generative language model (trained on
pseudo-text), and a speech decoder (generating a waveform from pseudo-text) all
trained without supervision and validate the proposed metrics with human
evaluation. Across 3 speech encoders (CPC, wav2vec 2.0, HuBERT), we find that
the number of discrete units (50, 100, or 200) matters in a task-dependent and
encoder-dependent way, and that some combinations approach text-based systems.
| 2,021 |
Computation and Language
|
"Is depression related to cannabis?": A knowledge-infused model for
Entity and Relation Extraction with Limited Supervision
|
With strong marketing advocacy of the benefits of cannabis use for improved
mental health, cannabis legalization is a priority among legislators. However,
preliminary scientific research does not conclusively associate cannabis with
improved mental health. In this study, we explore the relationship between
depression and consumption of cannabis in a targeted social media corpus
involving personal use of cannabis with the intent to derive its potential
mental health benefit. We use tweets that contain an association among three
categories annotated by domain experts - Reason, Effect, and Addiction. The
state-of-the-art Natural Langauge Processing techniques fall short in
extracting these relationships between cannabis phrases and the depression
indicators. We seek to address the limitation by using domain knowledge;
specifically, the Drug Abuse Ontology for addiction augmented with Diagnostic
and Statistical Manual of Mental Disorders lexicons for mental health. Because
of the lack of annotations due to the limited availability of the domain
experts' time, we use supervised contrastive learning in conjunction with GPT-3
trained on a vast corpus to achieve improved performance even with limited
supervision. Experimental results show that our method can significantly
extract cannabis-depression relationships better than the state-of-the-art
relation extractor. High-quality annotations can be provided using a nearest
neighbor approach using the learned representations that can be used by the
scientific community to understand the association between cannabis and
depression better.
| 2,021 |
Computation and Language
|
Inducing Meaningful Units from Character Sequences with Dynamic Capacity
Slot Attention
|
Characters do not convey meaning, but sequences of characters do. We propose
an unsupervised distributional method to learn the abstract meaningful units in
a sequence of characters. Rather than segmenting the sequence, our Dynamic
Capacity Slot Attention model discovers continuous representations of the
objects in the sequence, extending an architecture for object discovery in
images. We train our model on different languages and evaluate the quality of
the obtained representations with forward and reverse probing classifiers.
These experiments show that our model succeeds in discovering units which are
similar to those proposed previously in form, content and level of abstraction,
and which show promise for capturing meaningful information at a higher level
of abstraction.
| 2,024 |
Computation and Language
|
Self-Teaching Machines to Read and Comprehend with Large-Scale
Multi-Subject Question-Answering Data
|
In spite of much recent research in the area, it is still unclear whether
subject-area question-answering data is useful for machine reading
comprehension (MRC) tasks. In this paper, we investigate this question. We
collect a large-scale multi-subject multiple-choice question-answering dataset,
ExamQA, and use incomplete and noisy snippets returned by a web search engine
as the relevant context for each question-answering instance to convert it into
a weakly-labeled MRC instance. We then propose a self-teaching paradigm to
better use the generated weakly-labeled MRC instances to improve a target MRC
task. Experimental results show that we can obtain +5.1% in accuracy on a
multiple-choice MRC dataset, C^3, and +3.8% in exact match on an extractive MRC
dataset, CMRC 2018 over state-of-the-art MRC baselines, demonstrating the
effectiveness of our framework and the usefulness of large-scale subject-area
question-answering data for different types of machine reading comprehension
tasks.
| 2,021 |
Computation and Language
|
The impact of external innovation on new drug approvals: A retrospective
analysis
|
Pharmaceutical companies are relying more often on external sources of
innovation to boost their discovery research productivity. However, more
in-depth knowledge about how external innovation may translate to successful
product launches is still required in order to better understand how to best
leverage the innovation ecosystem. We analyzed the pre-approval publication
histories for FDA-approved new molecular entities (NMEs) and new biologic
entities (NBEs) launched by 13 top research pharma companies during the last
decade (2006-2016). We found that academic institutions contributed the
majority of pre-approval publications and that publication subject matter is
closely aligned with the strengths of the respective innovator. We found this
to also be true for candidate drugs terminated in Phase 3, but the volume of
literature on these molecules is substantially less than for approved drugs.
This may suggest that approved drugs are often associated with a more robust
dataset provided by a large number of institutes. Collectively, the results of
our analysis support the hypothesis that a collaborative research innovation
environment spanning across academia, industry and government is highly
conducive to successful drug approvals.
| 2,019 |
Computation and Language
|
MultiTalk: A Highly-Branching Dialog Testbed for Diverse Conversations
|
We study conversational dialog in which there are many possible responses to
a given history. We present the MultiTalk Dataset, a corpus of over 320,000
sentences of written conversational dialog that balances a high branching
factor (10) with several conversation turns (6) through selective branch
continuation. We make multiple contributions to study dialog generation in the
highly branching setting. In order to evaluate a diverse set of generations, we
propose a simple scoring algorithm, based on bipartite graph matching, to
optimally incorporate a set of diverse references. We study multiple language
generation tasks at different levels of predictive conversation depth, using
textual attributes induced automatically from pretrained classifiers. Our
culminating task is a challenging theory of mind problem, a controllable
generation task which requires reasoning about the expected reaction of the
listener.
| 2,021 |
Computation and Language
|
Neural Data Augmentation via Example Extrapolation
|
In many applications of machine learning, certain categories of examples may
be underrepresented in the training data, causing systems to underperform on
such "few-shot" cases at test time. A common remedy is to perform data
augmentation, such as by duplicating underrepresented examples, or
heuristically synthesizing new examples. But these remedies often fail to cover
the full diversity and complexity of real examples.
We propose a data augmentation approach that performs neural Example
Extrapolation (Ex2). Given a handful of exemplars sampled from some
distribution, Ex2 synthesizes new examples that also belong to the same
distribution. The Ex2 model is learned by simulating the example generation
procedure on data-rich slices of the data, and it is applied to
underrepresented, few-shot slices.
We apply Ex2 to a range of language understanding tasks and significantly
improve over state-of-the-art methods on multiple few-shot learning benchmarks,
including for relation extraction (FewRel) and intent classification + slot
filling (SNIPS).
| 2,021 |
Computation and Language
|
An Improved Baseline for Sentence-level Relation Extraction
|
Sentence-level relation extraction (RE) aims at identifying the relationship
between two entities in a sentence. Many efforts have been devoted to this
problem, while the best performing methods are still far from perfect. In this
paper, we revisit two problems that affect the performance of existing RE
models, namely entity representation and noisy or ill-defined labels. Our
improved RE baseline, incorporated with entity representations with typed
markers, achieves an F1 of 74.6% on TACRED, significantly outperforms previous
SOTA methods. Furthermore, the presented new baseline achieves an F1 of 91.1%
on the refined Re-TACRED dataset, demonstrating that the pretrained language
models (PLMs) achieve high performance on this task. We release our code to the
community for future research.
| 2,022 |
Computation and Language
|
Two Demonstrations of the Machine Translation Applications to Historical
Documents
|
We present our demonstration of two machine translation applications to
historical documents. The first task consists in generating a new version of a
historical document, written in the modern version of its original language.
The second application is limited to a document's orthography. It adapts the
document's spelling to modern standards in order to achieve an orthography
consistency and accounting for the lack of spelling conventions. We followed an
interactive, adaptive framework that allows the user to introduce corrections
to the system's hypothesis. The system reacts to these corrections by
generating a new hypothesis that takes them into account. Once the user is
satisfied with the system's hypothesis and validates it, the system adapts its
model following an online learning strategy. This system is implemented
following a client-server architecture. We developed a website which
communicates with the neural models. All code is open-source and publicly
available. The demonstration is hosted at http://demosmt.prhlt.upv.es/mthd/.
| 2,021 |
Computation and Language
|
MAUVE: Measuring the Gap Between Neural Text and Human Text using
Divergence Frontiers
|
As major progress is made in open-ended text generation, measuring how close
machine-generated text is to human language remains a critical open problem. We
introduce MAUVE, a comparison measure for open-ended text generation, which
directly compares the learnt distribution from a text generation model to the
distribution of human-written text using divergence frontiers. MAUVE scales up
to modern text generation models by computing information divergences in a
quantized embedding space. Through an extensive empirical study on three
open-ended generation tasks, we find that MAUVE identifies known properties of
generated text, scales naturally with model size, and correlates with human
judgments, with fewer restrictions than existing distributional evaluation
metrics.
| 2,021 |
Computation and Language
|
Clickbait Headline Detection in Indonesian News Sites using Multilingual
Bidirectional Encoder Representations from Transformers (M-BERT)
|
Click counts are related to the amount of money that online advertisers paid
to news sites. Such business models forced some news sites to employ a dirty
trick of click-baiting, i.e., using a hyperbolic and interesting words,
sometimes unfinished sentence in a headline to purposefully tease the readers.
Some Indonesian online news sites also joined the party of clickbait, which
indirectly degrade other established news sites' credibility. A neural network
with a pre-trained language model M-BERT that acted as a embedding layer is
then combined with a 100 nodes hidden layer and topped with a sigmoid
classifier was trained to detect clickbait headlines. With a total of 6632
headlines as a training dataset, the classifier performed remarkably well.
Evaluated with 5-fold cross validation, it has an accuracy score of 0.914,
f1-score of 0.914, precision score of 0.916, and ROC-AUC of 0.92. The usage of
multilingual BERT in Indonesian text classification task was tested and is
possible to be enhanced further. Future possibilities, societal impact, and
limitations of the clickbait detection are discussed.
| 2,021 |
Computation and Language
|
On Robustness of Neural Semantic Parsers
|
Semantic parsing maps natural language (NL) utterances into logical forms
(LFs), which underpins many advanced NLP problems. Semantic parsers gain
performance boosts with deep neural networks, but inherit vulnerabilities
against adversarial examples. In this paper, we provide the empirical study on
the robustness of semantic parsers in the presence of adversarial attacks.
Formally, adversaries of semantic parsing are considered to be the perturbed
utterance-LF pairs, whose utterances have exactly the same meanings as the
original ones. A scalable methodology is proposed to construct robustness test
sets based on existing benchmark corpora. Our results answered five research
questions in measuring the sate-of-the-art parsers' performance on robustness
test sets, and evaluating the effect of data augmentation.
| 2,021 |
Computation and Language
|
CTC-based Compression for Direct Speech Translation
|
Previous studies demonstrated that a dynamic phone-informed compression of
the input audio is beneficial for speech translation (ST). However, they
required a dedicated model for phone recognition and did not test this solution
for direct ST, in which a single model translates the input audio into the
target language without intermediate representations. In this work, we propose
the first method able to perform a dynamic compression of the input indirect ST
models. In particular, we exploit the Connectionist Temporal Classification
(CTC) to compress the input sequence according to its phonetic characteristics.
Our experiments demonstrate that our solution brings a 1.3-1.5 BLEU improvement
over a strong baseline on two language pairs (English-Italian and
English-German), contextually reducing the memory footprint by more than 10%.
| 2,021 |
Computation and Language
|
The GEM Benchmark: Natural Language Generation, its Evaluation and
Metrics
|
We introduce GEM, a living benchmark for natural language Generation (NLG),
its Evaluation, and Metrics. Measuring progress in NLG relies on a constantly
evolving ecosystem of automated metrics, datasets, and human evaluation
standards. Due to this moving target, new models often still evaluate on
divergent anglo-centric corpora with well-established, but flawed, metrics.
This disconnect makes it challenging to identify the limitations of current
models and opportunities for progress. Addressing this limitation, GEM provides
an environment in which models can easily be applied to a wide set of tasks and
in which evaluation strategies can be tested. Regular updates to the benchmark
will help NLG research become more multilingual and evolve the challenge
alongside models. This paper serves as the description of the data for which we
are organizing a shared task at our ACL 2021 Workshop and to which we invite
the entire NLG community to participate.
| 2,021 |
Computation and Language
|
The Multilingual TEDx Corpus for Speech Recognition and Translation
|
We present the Multilingual TEDx corpus, built to support speech recognition
(ASR) and speech translation (ST) research across many non-English source
languages. The corpus is a collection of audio recordings from TEDx talks in 8
source languages. We segment transcripts into sentences and align them to the
source-language audio and target-language translations. The corpus is released
along with open-sourced code enabling extension to new talks and languages as
they become available. Our corpus creation methodology can be applied to more
languages than previous work, and creates multi-way parallel evaluation sets.
We provide baselines in multiple ASR and ST settings, including multilingual
models to improve translation performance for low-resource language pairs.
| 2,021 |
Computation and Language
|
Memorization vs. Generalization: Quantifying Data Leakage in NLP
Performance Evaluation
|
Public datasets are often used to evaluate the efficacy and generalizability
of state-of-the-art methods for many tasks in natural language processing
(NLP). However, the presence of overlap between the train and test datasets can
lead to inflated results, inadvertently evaluating the model's ability to
memorize and interpreting it as the ability to generalize. In addition, such
data sets may not provide an effective indicator of the performance of these
methods in real world scenarios. We identify leakage of training data into test
data on several publicly available datasets used to evaluate NLP tasks,
including named entity recognition and relation extraction, and study them to
assess the impact of that leakage on the model's ability to memorize versus
generalize.
| 2,021 |
Computation and Language
|
A Computational Framework for Slang Generation
|
Slang is a common type of informal language, but its flexible nature and
paucity of data resources present challenges for existing natural language
systems. We take an initial step toward machine generation of slang by
developing a framework that models the speaker's word choice in slang context.
Our framework encodes novel slang meaning by relating the conventional and
slang senses of a word while incorporating syntactic and contextual knowledge
in slang usage. We construct the framework using a combination of probabilistic
inference and neural contrastive learning. We perform rigorous evaluations on
three slang dictionaries and show that our approach not only outperforms
state-of-the-art language models, but also better predicts the historical
emergence of slang word usages from 1960s to 2000s. We interpret the proposed
models and find that the contrastively learned semantic space is sensitive to
the similarities between slang and conventional senses of words. Our work
creates opportunities for the automated generation and interpretation of
informal language.
| 2,021 |
Computation and Language
|
An Investigation Between Schema Linking and Text-to-SQL Performance
|
Text-to-SQL is a crucial task toward developing methods for understanding
natural language by computers. Recent neural approaches deliver excellent
performance; however, models that are difficult to interpret inhibit future
developments. Hence, this study aims to provide a better approach toward the
interpretation of neural models. We hypothesize that the internal behavior of
models at hand becomes much easier to analyze if we identify the detailed
performance of schema linking simultaneously as the additional information of
the text-to-SQL performance. We provide the ground-truth annotation of schema
linking information onto the Spider dataset. We demonstrate the usefulness of
the annotated data and how to analyze the current state-of-the-art neural
models.
| 2,021 |
Computation and Language
|
HeBERT & HebEMO: a Hebrew BERT Model and a Tool for Polarity Analysis
and Emotion Recognition
|
This paper introduces HeBERT and HebEMO. HeBERT is a Transformer-based model
for modern Hebrew text, which relies on a BERT (Bidirectional Encoder
Representations for Transformers) architecture. BERT has been shown to
outperform alternative architectures in sentiment analysis, and is suggested to
be particularly appropriate for MRLs. Analyzing multiple BERT specifications,
we find that while model complexity correlates with high performance on
language tasks that aim to understand terms in a sentence, a more-parsimonious
model better captures the sentiment of entire sentence. Either way, out
BERT-based language model outperforms all existing Hebrew alternatives on all
common language tasks. HebEMO is a tool that uses HeBERT to detect polarity and
extract emotions from Hebrew UGC. HebEMO is trained on a unique
Covid-19-related UGC dataset that we collected and annotated for this study.
Data collection and annotation followed an active learning procedure that aimed
to maximize predictability. We show that HebEMO yields a high F1-score of 0.96
for polarity classification. Emotion detection reaches F1-scores of 0.78-0.97
for various target emotions, with the exception of surprise, which the model
failed to capture (F1 = 0.41). These results are better than the best-reported
performance, even among English-language models of emotion detection.
| 2,022 |
Computation and Language
|
Mind the Gap: Assessing Temporal Generalization in Neural Language
Models
|
Our world is open-ended, non-stationary, and constantly evolving; thus what
we talk about and how we talk about it change over time. This inherent dynamic
nature of language contrasts with the current static language modelling
paradigm, which trains and evaluates models on utterances from overlapping time
periods. Despite impressive recent progress, we demonstrate that Transformer-XL
language models perform worse in the realistic setup of predicting future
utterances from beyond their training period, and that model performance
becomes increasingly worse with time. We find that, while increasing model size
alone -- a key driver behind recent progress -- does not solve this problem,
having models that continually update their knowledge with new information can
indeed mitigate this performance degradation over time. Hence, given the
compilation of ever-larger language modelling datasets, combined with the
growing list of language-model-based NLP applications that require up-to-date
factual knowledge about the world, we argue that now is the right time to
rethink the static way in which we currently train and evaluate our language
models, and develop adaptive language models that can remain up-to-date with
respect to our ever-changing and non-stationary world. We publicly release our
dynamic, streaming language modelling benchmarks for WMT and arXiv to
facilitate language model evaluation that takes temporal dynamics into account.
| 2,021 |
Computation and Language
|
Top-down Discourse Parsing via Sequence Labelling
|
We introduce a top-down approach to discourse parsing that is conceptually
simpler than its predecessors (Kobayashi et al., 2020; Zhang et al., 2020). By
framing the task as a sequence labelling problem where the goal is to
iteratively segment a document into individual discourse units, we are able to
eliminate the decoder and reduce the search space for splitting points. We
explore both traditional recurrent models and modern pre-trained transformer
models for the task, and additionally introduce a novel dynamic oracle for
top-down parsing. Based on the Full metric, our proposed LSTM model sets a new
state-of-the-art for RST parsing.
| 2,021 |
Computation and Language
|
Learning to Select External Knowledge with Multi-Scale Negative Sampling
|
The Track-1 of DSTC9 aims to effectively answer user requests or questions
during task-oriented dialogues, which are out of the scope of APIs/DB. By
leveraging external knowledge resources, relevant information can be retrieved
and encoded into the response generation for these out-of-API-coverage queries.
In this work, we have explored several advanced techniques to enhance the
utilization of external knowledge and boost the quality of response generation,
including schema guided knowledge decision, negatives enhanced knowledge
selection, and knowledge grounded response generation. To evaluate the
performance of our proposed method, comprehensive experiments have been carried
out on the publicly available dataset. Our approach was ranked as the best in
human evaluation of DSTC9 Track-1.
| 2,021 |
Computation and Language
|
Learning to Match Mathematical Statements with Proofs
|
We introduce a novel task consisting in assigning a proof to a given
mathematical statement. The task is designed to improve the processing of
research-level mathematical texts. Applying Natural Language Processing (NLP)
tools to research level mathematical articles is both challenging, since it is
a highly specialized domain which mixes natural language and mathematical
formulae. It is also an important requirement for developing tools for
mathematical information retrieval and computer-assisted theorem proving. We
release a dataset for the task, consisting of over 180k statement-proof pairs
extracted from mathematical research articles. We carry out preliminary
experiments to assess the difficulty of the task. We first experiment with two
bag-of-words baselines. We show that considering the assignment problem
globally and using weighted bipartite matching algorithms helps a lot in
tackling the task. Finally, we introduce a self-attention-based model that can
be trained either locally or globally and outperforms baselines by a wide
margin.
| 2,021 |
Computation and Language
|
Introduction to Neural Transfer Learning with Transformers for Social
Science Text Analysis
|
Transformer-based models for transfer learning have the potential to achieve
high prediction accuracies on text-based supervised learning tasks with
relatively few training data instances. These models are thus likely to benefit
social scientists that seek to have as accurate as possible text-based measures
but only have limited resources for annotating training data. To enable social
scientists to leverage these potential benefits for their research, this paper
explains how these methods work, why they might be advantageous, and what their
limitations are. Additionally, three Transformer-based models for transfer
learning, BERT (Devlin et al. 2019), RoBERTa (Liu et al. 2019), and the
Longformer (Beltagy et al. 2020), are compared to conventional machine learning
algorithms on three applications. Across all evaluated tasks, textual styles,
and training data set sizes, the conventional models are consistently
outperformed by transfer learning with Transformers, thereby demonstrating the
benefits these models can bring to text-based social science research.
| 2,022 |
Computation and Language
|
Detecting Bias in Transfer Learning Approaches for Text Classification
|
Classification is an essential and fundamental task in machine learning,
playing a cardinal role in the field of natural language processing (NLP) and
computer vision (CV). In a supervised learning setting, labels are always
needed for the classification task. Especially for deep neural models, a large
amount of high-quality labeled data are required for training. However, when a
new domain comes out, it is usually hard or expensive to acquire the labels.
Transfer learning could be an option to transfer the knowledge from a source
domain to a target domain. A challenge is that these two domains can be
different, either on the feature distribution, or the class distribution for
the nature of the samples. In this work, we evaluate some existing transfer
learning approaches on detecting the bias of imbalanced classes including
traditional and deep models. Besides, we propose an approach to bridge the gap
of the domain class imbalance issue.
| 2,021 |
Computation and Language
|
Disambiguatory Signals are Stronger in Word-initial Positions
|
Psycholinguistic studies of human word processing and lexical access provide
ample evidence of the preferred nature of word-initial versus word-final
segments, e.g., in terms of attention paid by listeners (greater) or the
likelihood of reduction by speakers (lower). This has led to the conjecture --
as in Wedel et al. (2019b), but common elsewhere -- that languages have evolved
to provide more information earlier in words than later. Information-theoretic
methods to establish such tendencies in lexicons have suffered from several
methodological shortcomings that leave open the question of whether this high
word-initial informativeness is actually a property of the lexicon or simply an
artefact of the incremental nature of recognition. In this paper, we point out
the confounds in existing methods for comparing the informativeness of segments
early in the word versus later in the word, and present several new measures
that avoid these confounds. When controlling for these confounds, we still find
evidence across hundreds of languages that indeed there is a cross-linguistic
tendency to front-load information in words.
| 2,021 |
Computation and Language
|
Bootstrapping Multilingual AMR with Contextual Word Alignments
|
We develop high performance multilingualAbstract Meaning Representation (AMR)
sys-tems by projecting English AMR annotationsto other languages with weak
supervision. Weachieve this goal by bootstrapping transformer-based
multilingual word embeddings, in partic-ular those from cross-lingual RoBERTa
(XLM-R large). We develop a novel technique forforeign-text-to-English AMR
alignment, usingthe contextual word alignment between En-glish and foreign
language tokens. This wordalignment is weakly supervised and relies onthe
contextualized XLM-R word embeddings.We achieve a highly competitive
performancethat surpasses the best published results forGerman, Italian,
Spanish and Chinese.
| 2,021 |
Computation and Language
|
DiSCoL: Toward Engaging Dialogue Systems through Conversational Line
Guided Response Generation
|
Having engaging and informative conversations with users is the utmost goal
for open-domain conversational systems. Recent advances in transformer-based
language models and their applications to dialogue systems have succeeded to
generate fluent and human-like responses. However, they still lack control over
the generation process towards producing contentful responses and achieving
engaging conversations. To achieve this goal, we present \textbf{DiSCoL}
(\textbf{Di}alogue \textbf{S}ystems through \textbf{Co}versational
\textbf{L}ine guided response generation). DiSCoL is an open-domain dialogue
system that leverages conversational lines (briefly \textbf{convlines}) as
controllable and informative content-planning elements to guide the generation
model produce engaging and informative responses. Two primary modules in
DiSCoL's pipeline are conditional generators trained for 1) predicting relevant
and informative convlines for dialogue contexts and 2) generating high-quality
responses conditioned on the predicted convlines. Users can also change the
returned convlines to \textit{control} the direction of the conversations
towards topics that are more interesting for them. Through automatic and human
evaluations, we demonstrate the efficiency of the convlines in producing
engaging conversations.
| 2,021 |
Computation and Language
|
When Can Models Learn From Explanations? A Formal Framework for
Understanding the Roles of Explanation Data
|
Many methods now exist for conditioning model outputs on task instructions,
retrieved documents, and user-provided explanations and feedback. Rather than
relying solely on examples of task inputs and outputs, these approaches use
valuable additional data for improving model correctness and aligning learned
models with human priors. Meanwhile, a growing body of evidence suggests that
some language models can (1) store a large amount of knowledge in their
parameters, and (2) perform inference over tasks in textual inputs at test
time. These results raise the possibility that, for some tasks, humans cannot
explain to a model any more about the task than it already knows or could infer
on its own. In this paper, we study the circumstances under which explanations
of individual data points can (or cannot) improve modeling performance. In
order to carefully control important properties of the data and explanations,
we introduce a synthetic dataset for experiments, and we also make use of three
existing datasets with explanations: e-SNLI, TACRED, and SemEval. We first give
a formal framework for the available modeling approaches, in which explanation
data can be used as model inputs, as targets, or as a prior. After arguing that
the most promising role for explanation data is as model inputs, we propose to
use a retrieval-based method and show that it solves our synthetic task with
accuracies upwards of 95%, while baselines without explanation data achieve
below 65% accuracy. We then identify properties of datasets for which
retrieval-based modeling fails. With the three existing datasets, we find no
improvements from explanation retrieval. Drawing on findings from our synthetic
task, we suggest that at least one of six preconditions for successful modeling
fails to hold with these datasets. Our code is publicly available at
https://github.com/peterbhase/ExplanationRoles
| 2,021 |
Computation and Language
|
Confusion2vec 2.0: Enriching Ambiguous Spoken Language Representations
with Subwords
|
Word vector representations enable machines to encode human language for
spoken language understanding and processing. Confusion2vec, motivated from
human speech production and perception, is a word vector representation which
encodes ambiguities present in human spoken language in addition to semantics
and syntactic information. Confusion2vec provides a robust spoken language
representation by considering inherent human language ambiguities. In this
paper, we propose a novel word vector space estimation by unsupervised learning
on lattices output by an automatic speech recognition (ASR) system. We encode
each word in confusion2vec vector space by its constituent subword character
n-grams. We show the subword encoding helps better represent the acoustic
perceptual ambiguities in human spoken language via information modeled on
lattice structured ASR output. The usefulness of the proposed Confusion2vec
representation is evaluated using semantic, syntactic and acoustic analogy and
word similarity tasks. We also show the benefits of subword modeling for
acoustic ambiguity representation on the task of spoken language intent
detection. The results significantly outperform existing word vector
representations when evaluated on erroneous ASR outputs. We demonstrate that
Confusion2vec subword modeling eliminates the need for retraining/adapting the
natural language understanding models on ASR transcripts.
| 2,022 |
Computation and Language
|
Self-Supervised Claim Identification for Automated Fact Checking
|
We propose a novel, attention-based self-supervised approach to identify
"claim-worthy" sentences in a fake news article, an important first step in
automated fact-checking. We leverage "aboutness" of headline and content using
attention mechanism for this task. The identified claims can be used for
downstream task of claim verification for which we are releasing a benchmark
dataset of manually selected compelling articles with veracity labels and
associated evidence. This work goes beyond stylistic analysis to identifying
content that influences reader belief. Experiments with three datasets show the
strength of our model. Data and code available at
https://github.com/architapathak/Self-Supervised-ClaimIdentification
| 2,021 |
Computation and Language
|
Converse, Focus and Guess -- Towards Multi-Document Driven Dialogue
|
We propose a novel task, Multi-Document Driven Dialogue (MD3), in which an
agent can guess the target document that the user is interested in by leading a
dialogue. To benchmark progress, we introduce a new dataset of GuessMovie,
which contains 16,881 documents, each describing a movie, and associated 13,434
dialogues. Further, we propose the MD3 model. Keeping guessing the target
document in mind, it converses with the user conditioned on both document
engagement and user feedback. In order to incorporate large-scale external
documents into the dialogue, it pretrains a document representation which is
sensitive to attributes it talks about an object. Then it tracks dialogue state
by detecting evolvement of document belief and attribute belief, and finally
optimizes dialogue policy in principle of entropy decreasing and reward
increasing, which is expected to successfully guess the user's target in a
minimum number of turns. Experiments show that our method significantly
outperforms several strong baseline methods and is very close to human's
performance.
| 2,021 |
Computation and Language
|
Bangla Text Dataset and Exploratory Analysis for Online Harassment
Detection
|
Being the seventh most spoken language in the world, the use of the Bangla
language online has increased in recent times. Hence, it has become very
important to analyze Bangla text data to maintain a safe and harassment-free
online place. The data that has been made accessible in this article has been
gathered and marked from the comments of people in public posts by celebrities,
government officials, athletes on Facebook. The total amount of collected
comments is 44001. The dataset is compiled with the aim of developing the
ability of machines to differentiate whether a comment is a bully expression or
not with the help of Natural Language Processing and to what extent it is
improper if it is an inappropriate comment. The comments are labeled with
different categories of harassment. Exploratory analysis from different
perspectives is also included in this paper to have a detailed overview. Due to
the scarcity of data collection of categorized Bengali language comments, this
dataset can have a significant role for research in detecting bully words,
identifying inappropriate comments, detecting different categories of Bengali
bullies, etc. The dataset is publicly available at
https://data.mendeley.com/datasets/9xjx8twk8p.
| 2,021 |
Computation and Language
|
Understanding the Capabilities, Limitations, and Societal Impact of
Large Language Models
|
On October 14th, 2020, researchers from OpenAI, the Stanford Institute for
Human-Centered Artificial Intelligence, and other universities convened to
discuss open research questions surrounding GPT-3, the largest
publicly-disclosed dense language model at the time. The meeting took place
under Chatham House Rules. Discussants came from a variety of research
backgrounds including computer science, linguistics, philosophy, political
science, communications, cyber policy, and more. Broadly, the discussion
centered around two main questions: 1) What are the technical capabilities and
limitations of large language models? 2) What are the societal effects of
widespread use of large language models? Here, we provide a detailed summary of
the discussion organized by the two themes above.
| 2,021 |
Computation and Language
|
Adaptive Semiparametric Language Models
|
We present a language model that combines a large parametric neural network
(i.e., a transformer) with a non-parametric episodic memory component in an
integrated architecture. Our model uses extended short-term context by caching
local hidden states -- similar to transformer-XL -- and global long-term memory
by retrieving a set of nearest neighbor tokens at each timestep. We design a
gating function to adaptively combine multiple information sources to make a
prediction. This mechanism allows the model to use either local context,
short-term memory, or long-term memory (or any combination of them) on an ad
hoc basis depending on the context. Experiments on word-based and
character-based language modeling datasets demonstrate the efficacy of our
proposed method compared to strong baselines.
| 2,021 |
Computation and Language
|
Incremental Beam Manipulation for Natural Language Generation
|
The performance of natural language generation systems has improved
substantially with modern neural networks. At test time they typically employ
beam search to avoid locally optimal but globally suboptimal predictions.
However, due to model errors, a larger beam size can lead to deteriorating
performance according to the evaluation metric. For this reason, it is common
to rerank the output of beam search, but this relies on beam search to produce
a good set of hypotheses, which limits the potential gains. Other alternatives
to beam search require changes to the training of the model, which restricts
their applicability compared to beam search. This paper proposes incremental
beam manipulation, i.e. reranking the hypotheses in the beam during decoding
instead of only at the end. This way, hypotheses that are unlikely to lead to a
good final output are discarded, and in their place hypotheses that would have
been ignored will be considered instead. Applying incremental beam manipulation
leads to an improvement of 1.93 and 5.82 BLEU points over vanilla beam search
for the test sets of the E2E and WebNLG challenges respectively. The proposed
method also outperformed a strong reranker by 1.04 BLEU points on the E2E
challenge, while being on par with it on the WebNLG dataset.
| 2,021 |
Computation and Language
|
One Size Does Not Fit All: Finding the Optimal Subword Sizes for
FastText Models across Languages
|
Unsupervised representation learning of words from large multilingual corpora
is useful for downstream tasks such as word sense disambiguation, semantic text
similarity, and information retrieval. The representation precision of
log-bilinear fastText models is mostly due to their use of subword information.
In previous work, the optimization of fastText's subword sizes has not been
fully explored, and non-English fastText models were trained using subword
sizes optimized for English and German word analogy tasks. In our work, we find
the optimal subword sizes on the English, German, Czech, Italian, Spanish,
French, Hindi, Turkish, and Russian word analogy tasks. We then propose a
simple n-gram coverage model and we show that it predicts better-than-default
subword sizes on the Spanish, French, Hindi, Turkish, and Russian word analogy
tasks. We show that the optimization of fastText's subword sizes matters and
results in a 14% improvement on the Czech word analogy task. We also show that
expensive parameter optimization can be replaced by a simple n-gram coverage
model that consistently improves the accuracy of fastText models on the word
analogy tasks by up to 3% compared to the default subword sizes, and that it is
within 1% accuracy of the optimal subword sizes.
| 2,021 |
Computation and Language
|
Data-to-text Generation with Macro Planning
|
Recent approaches to data-to-text generation have adopted the very successful
encoder-decoder architecture or variants thereof. These models generate text
which is fluent (but often imprecise) and perform quite poorly at selecting
appropriate content and ordering it coherently. To overcome some of these
issues, we propose a neural model with a macro planning stage followed by a
generation stage reminiscent of traditional methods which embrace separate
modules for planning and surface realization. Macro plans represent high level
organization of important content such as entities, events and their
interactions; they are learnt from data and given as input to the generator.
Extensive experiments on two data-to-text benchmarks (RotoWire and MLB) show
that our approach outperforms competitive baselines in terms of automatic and
human evaluation.
| 2,021 |
Computation and Language
|
Unifying Vision-and-Language Tasks via Text Generation
|
Existing methods for vision-and-language learning typically require designing
task-specific architectures and objectives for each task. For example, a
multi-label answer classifier for visual question answering, a region scorer
for referring expression comprehension, and a language decoder for image
captioning, etc. To alleviate these hassles, in this work, we propose a unified
framework that learns different tasks in a single architecture with the same
language modeling objective, i.e., multimodal conditional text generation,
where our models learn to generate labels in text based on the visual and
textual inputs. On 7 popular vision-and-language benchmarks, including visual
question answering, referring expression comprehension, visual commonsense
reasoning, most of which have been previously modeled as discriminative tasks,
our generative approach (with a single unified architecture) reaches comparable
performance to recent task-specific state-of-the-art vision-and-language
models. Moreover, our generative approach shows better generalization ability
on questions that have rare answers. Also, we show that our framework allows
multi-task learning in a single architecture with a single set of parameters,
achieving similar performance to separately optimized single-task models. Our
code is publicly available at: https://github.com/j-min/VL-T5
| 2,021 |
Computation and Language
|
Controlling Hallucinations at Word Level in Data-to-Text Generation
|
Data-to-Text Generation (DTG) is a subfield of Natural Language Generation
aiming at transcribing structured data in natural language descriptions. The
field has been recently boosted by the use of neural-based generators which
exhibit on one side great syntactic skills without the need of hand-crafted
pipelines; on the other side, the quality of the generated text reflects the
quality of the training data, which in realistic settings only offer
imperfectly aligned structure-text pairs. Consequently, state-of-art neural
models include misleading statements - usually called hallucinations - in their
outputs. The control of this phenomenon is today a major challenge for DTG, and
is the problem addressed in the paper.
Previous work deal with this issue at the instance level: using an alignment
score for each table-reference pair. In contrast, we propose a finer-grained
approach, arguing that hallucinations should rather be treated at the word
level. Specifically, we propose a Multi-Branch Decoder which is able to
leverage word-level labels to learn the relevant parts of each training
instance. These labels are obtained following a simple and efficient scoring
procedure based on co-occurrence analysis and dependency parsing. Extensive
evaluations, via automated metrics and human judgment on the standard WikiBio
benchmark, show the accuracy of our alignment labels and the effectiveness of
the proposed Multi-Branch Decoder. Our model is able to reduce and control
hallucinations, while keeping fluency and coherence in generated texts. Further
experiments on a degraded version of ToTTo show that our model could be
successfully used on very noisy settings.
| 2,021 |
Computation and Language
|
Building Representative Corpora from Illiterate Communities: A Review of
Challenges and Mitigation Strategies for Developing Countries
|
Most well-established data collection methods currently adopted in NLP depend
on the assumption of speaker literacy. Consequently, the collected corpora
largely fail to represent swathes of the global population, which tend to be
some of the most vulnerable and marginalised people in society, and often live
in rural developing areas. Such underrepresented groups are thus not only
ignored when making modeling and system design decisions, but also prevented
from benefiting from development outcomes achieved through data-driven NLP.
This paper aims to address the under-representation of illiterate communities
in NLP corpora: we identify potential biases and ethical issues that might
arise when collecting data from rural communities with high illiteracy rates in
Low-Income Countries, and propose a set of practical mitigation strategies to
help future work.
| 2,021 |
Computation and Language
|
Generalized Zero-shot Intent Detection via Commonsense Knowledge
|
Identifying user intents from natural language utterances is a crucial step
in conversational systems that has been extensively studied as a supervised
classification problem. However, in practice, new intents emerge after
deploying an intent detection model. Thus, these models should seamlessly adapt
and classify utterances with both seen and unseen intents -- unseen intents
emerge after deployment and they do not have training data. The few existing
models that target this setting rely heavily on the scarcely available training
data and overfit to seen intents data, resulting in a bias to misclassify
utterances with unseen intents into seen ones. We propose RIDE: an intent
detection model that leverages commonsense knowledge in an unsupervised fashion
to overcome the issue of training data scarcity. RIDE computes robust and
generalizable relationship meta-features that capture deep semantic
relationships between utterances and intent labels; these features are computed
by considering how the concepts in an utterance are linked to those in an
intent label via commonsense knowledge. Our extensive experimental analysis on
three widely-used intent detection benchmarks shows that relationship
meta-features significantly increase the accuracy of detecting both seen and
unseen intents and that RIDE outperforms the state-of-the-art model for unseen
intents.
| 2,021 |
Computation and Language
|
Understanding Pre-Editing for Black-Box Neural Machine Translation
|
Pre-editing is the process of modifying the source text (ST) so that it can
be translated by machine translation (MT) in a better quality. Despite the
unpredictability of black-box neural MT (NMT), pre-editing has been deployed in
various practical MT use cases. Although many studies have demonstrated the
effectiveness of pre-editing methods for particular settings, thus far, a deep
understanding of what pre-editing is and how it works for black-box NMT is
lacking. To elicit such understanding, we extensively investigated human
pre-editing practices. We first implemented a protocol to incrementally record
the minimum edits for each ST and collected 6,652 instances of pre-editing
across three translation directions, two MT systems, and four text domains. We
then analysed the instances from three perspectives: the characteristics of the
pre-edited ST, the diversity of pre-editing operations, and the impact of the
pre-editing operations on NMT outputs. Our findings include the following: (1)
enhancing the explicitness of the meaning of an ST and its syntactic structure
is more important for obtaining better translations than making the ST shorter
and simpler, and (2) although the impact of pre-editing on NMT is generally
unpredictable, there are some tendencies of changes in the NMT outputs
depending on the editing operation types.
| 2,021 |
Computation and Language
|
RpBERT: A Text-image Relation Propagation-based BERT Model for
Multimodal NER
|
Recently multimodal named entity recognition (MNER) has utilized images to
improve the accuracy of NER in tweets. However, most of the multimodal methods
use attention mechanisms to extract visual clues regardless of whether the text
and image are relevant. Practically, the irrelevant text-image pairs account
for a large proportion in tweets. The visual clues that are unrelated to the
texts will exert uncertain or even negative effects on multimodal model
learning. In this paper, we introduce a method of text-image relation
propagation into the multimodal BERT model. We integrate soft or hard gates to
select visual clues and propose a multitask algorithm to train on the MNER
datasets. In the experiments, we deeply analyze the changes in visual attention
before and after the use of text-image relation propagation. Our model achieves
state-of-the-art performance on the MNER datasets.
| 2,021 |
Computation and Language
|
GraphPlan: Story Generation by Planning with Event Graph
|
Story generation is a task that aims to automatically produce multiple
sentences to make up a meaningful story. This task is challenging because it
requires high-level understanding of semantic meaning of sentences and
causality of story events. Naive sequence-to-sequence models generally fail to
acquire such knowledge, as the logical correctness can hardly be guaranteed in
a text generation model without the strategic planning. In this paper, we focus
on planning a sequence of events assisted by event graphs, and use the events
to guide the generator. Instead of using a sequence-to-sequence model to output
a storyline as in some existing works, we propose to generate an event sequence
by walking on an event graph. The event graphs are built automatically based on
the corpus. To evaluate the proposed approach, we conduct human evaluation both
on event planning and story generation. Based on large-scale human annotation
results, our proposed approach is shown to produce more logically correct event
sequences and stories.
| 2,021 |
Computation and Language
|
Model Agnostic Answer Reranking System for Adversarial Question
Answering
|
While numerous methods have been proposed as defenses against adversarial
examples in question answering (QA), these techniques are often model specific,
require retraining of the model, and give only marginal improvements in
performance over vanilla models. In this work, we present a simple
model-agnostic approach to this problem that can be applied directly to any QA
model without any retraining. Our method employs an explicit answer candidate
reranking mechanism that scores candidate answers on the basis of their content
overlap with the question before making the final prediction. Combined with a
strong base QAmodel, our method outperforms state-of-the-art defense
techniques, calling into question how well these techniques are actually doing
and strong these adversarial testbeds are.
| 2,021 |
Computation and Language
|
Spell Correction for Azerbaijani Language using Deep Neural Networks
|
Spell correction is used to detect and correct orthographic mistakes in
texts. Most of the time, traditional dictionary lookup with string similarity
methods is suitable for the languages that have a less complex structure such
as the English language. However, the Azerbaijani language has a more complex
structure and due to its morphological structure, the derivation of words is
plenty that several words are derived from adding suffices, affixes to the
words. Therefore, in this paper sequence to sequence model with an attention
mechanism is used to develop spelling correction for Azerbaijani. Total 12000
wrong and correct sentence pairs used for training, and the model is tested on
1000 real-world misspelled words and F1-score results are 75% for distance 0,
90% for distance 1, and 96% for distance 2.
| 2,021 |
Computation and Language
|
Think you have Solved Direct-Answer Question Answering? Try ARC-DA, the
Direct-Answer AI2 Reasoning Challenge
|
We present the ARC-DA dataset, a direct-answer ("open response", "freeform")
version of the ARC (AI2 Reasoning Challenge) multiple-choice dataset. While ARC
has been influential in the community, its multiple-choice format is
unrepresentative of real-world questions, and multiple choice formats can be
particularly susceptible to artifacts. The ARC-DA dataset addresses these
concerns by converting questions to direct-answer format using a combination of
crowdsourcing and expert review. The resulting dataset contains 2985 questions
with a total of 8436 valid answers (questions typically have more than one
valid answer). ARC-DA is one of the first DA datasets of natural questions that
often require reasoning, and where appropriate question decompositions are not
evident from the questions themselves. We describe the conversion approach
taken, appropriate evaluation metrics, and several strong models. Although
high, the best scores (81% GENIE, 61.4% F1, 63.2% ROUGE-L) still leave
considerable room for improvement. In addition, the dataset provides a natural
setting for new research on explanation, as many questions require reasoning to
construct answers. We hope the dataset spurs further advances in complex
question-answering by the community. ARC-DA is available at
https://allenai.org/data/arc-da
| 2,021 |
Computation and Language
|
Child-directed Listening: How Caregiver Inference Enables Children's
Early Verbal Communication
|
How do adults understand children's speech? Children's productions over the
course of language development often bear little resemblance to typical adult
pronunciations, yet caregivers nonetheless reliably recover meaning from them.
Here, we employ a suite of Bayesian models of spoken word recognition to
understand how adults overcome the noisiness of child language, showing that
communicative success between children and adults relies heavily on adult
inferential processes. By evaluating competing models on phonetically-annotated
corpora, we show that adults' recovered meanings are best predicted by prior
expectations fitted specifically to the child language environment, rather than
to typical adult-adult language. After quantifying the contribution of this
"child-directed listening" over developmental time, we discuss the consequences
for theories of language acquisition, as well as the implications for
commonly-used methods for assessing children's linguistic proficiency.
| 2,021 |
Computation and Language
|
Jointly Improving Language Understanding and Generation with
Quality-Weighted Weak Supervision of Automatic Labeling
|
Neural natural language generation (NLG) and understanding (NLU) models are
data-hungry and require massive amounts of annotated data to be competitive.
Recent frameworks address this bottleneck with generative models that
synthesize weak labels at scale, where a small amount of training labels are
expert-curated and the rest of the data is automatically annotated. We follow
that approach, by automatically constructing a large-scale weakly-labeled data
with a fine-tuned GPT-2, and employ a semi-supervised framework to jointly
train the NLG and NLU models. The proposed framework adapts the parameter
updates to the models according to the estimated label-quality. On both the E2E
and Weather benchmarks, we show that this weakly supervised training paradigm
is an effective approach under low resource scenarios and outperforming
benchmark systems on both datasets when 100% of training data is used.
| 2,021 |
Computation and Language
|
Does the Order of Training Samples Matter? Improving Neural Data-to-Text
Generation with Curriculum Learning
|
Recent advancements in data-to-text generation largely take on the form of
neural end-to-end systems. Efforts have been dedicated to improving text
generation systems by changing the order of training samples in a process known
as curriculum learning. Past research on sequence-to-sequence learning showed
that curriculum learning helps to improve both the performance and convergence
speed. In this work, we delve into the same idea surrounding the training
samples consisting of structured data and text pairs, where at each update, the
curriculum framework selects training samples based on the model's competence.
Specifically, we experiment with various difficulty metrics and put forward a
soft edit distance metric for ranking training samples. Our benchmarks show
faster convergence speed where training time is reduced by 38.7% and
performance is boosted by 4.84 BLEU.
| 2,021 |
Computation and Language
|
Neural Data-to-Text Generation with LM-based Text Augmentation
|
For many new application domains for data-to-text generation, the main
obstacle in training neural models consists of a lack of training data. While
usually large numbers of instances are available on the data side, often only
very few text samples are available. To address this problem, we here propose a
novel few-shot approach for this setting. Our approach automatically augments
the data available for training by (i) generating new text samples based on
replacing specific values by alternative ones from the same category, (ii)
generating new text samples based on GPT-2, and (iii) proposing an automatic
method for pairing the new text samples with data samples. As the text
augmentation can introduce noise to the training data, we use cycle consistency
as an objective, in order to make sure that a given data sample can be
correctly reconstructed after having been formulated as text (and that text
samples can be reconstructed from data). On both the E2E and WebNLG benchmarks,
we show that this weakly supervised training paradigm is able to outperform
fully supervised seq2seq models with less than 10% annotations. By utilizing
all annotated data, our model can boost the performance of a standard seq2seq
model by over 5 BLEU points, establishing a new state-of-the-art on both
datasets.
| 2,021 |
Computation and Language
|
Does He Wink or Does He Nod? A Challenging Benchmark for Evaluating Word
Understanding of Language Models
|
Recent progress in pretraining language models on large corpora has resulted
in large performance gains on many NLP tasks. These large models acquire
linguistic knowledge during pretraining, which helps to improve performance on
downstream tasks via fine-tuning. To assess what kind of knowledge is acquired,
language models are commonly probed by querying them with `fill in the blank'
style cloze questions. Existing probing datasets mainly focus on knowledge
about relations between words and entities. We introduce WDLMPro (Word
Definition Language Model Probing) to evaluate word understanding directly
using dictionary definitions of words. In our experiments, three popular
pretrained language models struggle to match words and their definitions. This
indicates that they understand many words poorly and that our new probing task
is a difficult challenge that could help guide research on LMs in the future.
| 2,021 |
Computation and Language
|
A bandit approach to curriculum generation for automatic speech
recognition
|
The Automated Speech Recognition (ASR) task has been a challenging domain
especially for low data scenarios with few audio examples. This is the main
problem in training ASR systems on the data from low-resource or marginalized
languages. In this paper we present an approach to mitigate the lack of
training data by employing Automated Curriculum Learning in combination with an
adversarial bandit approach inspired by Reinforcement learning. The goal of the
approach is to optimize the training sequence of mini-batches ranked by the
level of difficulty and compare the ASR performance metrics against the random
training sequence and discrete curriculum. We test our approach on a truly
low-resource language and show that the bandit framework has a good improvement
over the baseline transfer-learning model.
| 2,021 |
Computation and Language
|
From Toxicity in Online Comments to Incivility in American News: Proceed
with Caution
|
The ability to quantify incivility online, in news and in congressional
debates, is of great interest to political scientists. Computational tools for
detecting online incivility for English are now fairly accessible and
potentially could be applied more broadly. We test the Jigsaw Perspective API
for its ability to detect the degree of incivility on a corpus that we
developed, consisting of manual annotations of civility in American news. We
demonstrate that toxicity models, as exemplified by Perspective, are inadequate
for the analysis of incivility in news. We carry out error analysis that points
to the need to develop methods to remove spurious correlations between words
often mentioned in the news, especially identity descriptors and incivility.
Without such improvements, applying Perspective or similar models on news is
likely to lead to wrong conclusions, that are not aligned with the human
perception of incivility.
| 2,021 |
Computation and Language
|
Representation Learning for Natural Language Processing
|
This book aims to review and present the recent advances of distributed
representation learning for NLP, including why representation learning can
improve NLP, how representation learning takes part in various important topics
of NLP, and what challenges are still not well addressed by distributed
representation.
| 2,021 |
Computation and Language
|
Memory Augmented Sequential Paragraph Retrieval for Multi-hop Question
Answering
|
Retrieving information from correlative paragraphs or documents to answer
open-domain multi-hop questions is very challenging. To deal with this
challenge, most of the existing works consider paragraphs as nodes in a graph
and propose graph-based methods to retrieve them. However, in this paper, we
point out the intrinsic defect of such methods. Instead, we propose a new
architecture that models paragraphs as sequential data and considers multi-hop
information retrieval as a kind of sequence labeling task. Specifically, we
design a rewritable external memory to model the dependency among paragraphs.
Moreover, a threshold gate mechanism is proposed to eliminate the distraction
of noise paragraphs. We evaluate our method on both full wiki and distractor
subtask of HotpotQA, a public textual multi-hop QA dataset requiring multi-hop
information retrieval. Experiments show that our method achieves significant
improvement over the published state-of-the-art method in retrieval and
downstream QA task performance.
| 2,021 |
Computation and Language
|
CSS-LM: A Contrastive Framework for Semi-supervised Fine-tuning of
Pre-trained Language Models
|
Fine-tuning pre-trained language models (PLMs) has demonstrated its
effectiveness on various downstream NLP tasks recently. However, in many
low-resource scenarios, the conventional fine-tuning strategies cannot
sufficiently capture the important semantic features for downstream tasks. To
address this issue, we introduce a novel framework (named "CSS-LM") to improve
the fine-tuning phase of PLMs via contrastive semi-supervised learning.
Specifically, given a specific task, we retrieve positive and negative
instances from large-scale unlabeled corpora according to their domain-level
and class-level semantic relatedness to the task. We then perform contrastive
semi-supervised learning on both the retrieved unlabeled and original labeled
instances to help PLMs capture crucial task-related semantic features. The
experimental results show that CSS-LM achieves better results than the
conventional fine-tuning strategy on a series of downstream tasks with few-shot
settings, and outperforms the latest supervised contrastive fine-tuning
strategies. Our datasets and source code will be available to provide more
details.
| 2,021 |
Computation and Language
|
Unsupervised Sentence-embeddings by Manifold Approximation and
Projection
|
The concept of unsupervised universal sentence encoders has gained traction
recently, wherein pre-trained models generate effective task-agnostic
fixed-dimensional representations for phrases, sentences and paragraphs. Such
methods are of varying complexity, from simple weighted-averages of word
vectors to complex language-models based on bidirectional transformers. In this
work we propose a novel technique to generate sentence-embeddings in an
unsupervised fashion by projecting the sentences onto a fixed-dimensional
manifold with the objective of preserving local neighbourhoods in the original
space. To delineate such neighbourhoods we experiment with several set-distance
metrics, including the recently proposed Word Mover's distance, while the
fixed-dimensional projection is achieved by employing a scalable and efficient
manifold approximation method rooted in topological data analysis. We test our
approach, which we term EMAP or Embeddings by Manifold Approximation and
Projection, on six publicly available text-classification datasets of varying
size and complexity. Empirical results show that our method consistently
performs similar to or better than several alternative state-of-the-art
approaches.
| 2,021 |
Computation and Language
|
Spoiler Alert: Using Natural Language Processing to Detect Spoilers in
Book Reviews
|
This paper presents an NLP (Natural Language Processing) approach to
detecting spoilers in book reviews, using the University of California San
Diego (UCSD) Goodreads Spoiler dataset. We explored the use of LSTM, BERT, and
RoBERTa language models to perform spoiler detection at the sentence-level.
This was contrasted with a UCSD paper which performed the same task, but using
handcrafted features in its data preparation. Despite eschewing the use of
handcrafted features, our results from the LSTM model were able to slightly
exceed the UCSD team's performance in spoiler detection.
| 2,021 |
Computation and Language
|
Nystr\"omformer: A Nystr\"om-Based Algorithm for Approximating
Self-Attention
|
Transformers have emerged as a powerful tool for a broad range of natural
language processing tasks. A key component that drives the impressive
performance of Transformers is the self-attention mechanism that encodes the
influence or dependence of other tokens on each specific token. While
beneficial, the quadratic complexity of self-attention on the input sequence
length has limited its application to longer sequences -- a topic being
actively studied in the community. To address this limitation, we propose
Nystr\"{o}mformer -- a model that exhibits favorable scalability as a function
of sequence length. Our idea is based on adapting the Nystr\"{o}m method to
approximate standard self-attention with $O(n)$ complexity. The scalability of
Nystr\"{o}mformer enables application to longer sequences with thousands of
tokens. We perform evaluations on multiple downstream tasks on the GLUE
benchmark and IMDB reviews with standard sequence length, and find that our
Nystr\"{o}mformer performs comparably, or in a few cases, even slightly better,
than standard self-attention. On longer sequence tasks in the Long Range Arena
(LRA) benchmark, Nystr\"{o}mformer performs favorably relative to other
efficient self-attention methods. Our code is available at
https://github.com/mlpen/Nystromformer.
| 2,021 |
Computation and Language
|
MirrorAlign: A Super Lightweight Unsupervised Word Alignment Model via
Cross-Lingual Contrastive Learning
|
Word alignment is essential for the downstream cross-lingual language
understanding and generation tasks. Recently, the performance of the neural
word alignment models has exceeded that of statistical models. However, they
heavily rely on sophisticated translation models. In this study, we propose a
super lightweight unsupervised word alignment model named MirrorAlign, in which
bidirectional symmetric attention trained with a contrastive learning objective
is introduced, and an agreement loss is employed to bind the attention maps,
such that the alignments follow mirror-like symmetry hypothesis. Experimental
results on several public benchmarks demonstrate that our model achieves
competitive, if not better, performance compared to the state of the art in
word alignment while significantly reducing the training and decoding time on
average. Further ablation analysis and case studies show the superiority of our
proposed MirrorAlign. Notably, we recognize our model as a pioneer attempt to
unify bilingual word embedding and word alignments. Encouragingly, our approach
achieves {16.4X speedup} against GIZA++, and {50X parameter compression}
compared with the Transformer-based alignment methods. We release our code to
facilitate the community: https://github.com/moore3930/MirrorAlign.
| 2,022 |
Computation and Language
|
Quality Estimation without Human-labeled Data
|
Quality estimation aims to measure the quality of translated content without
access to a reference translation. This is crucial for machine translation
systems in real-world scenarios where high-quality translation is needed. While
many approaches exist for quality estimation, they are based on supervised
machine learning requiring costly human labelled data. As an alternative, we
propose a technique that does not rely on examples from human-annotators and
instead uses synthetic training data. We train off-the-shelf architectures for
supervised quality estimation on our synthetic data and show that the resulting
models achieve comparable performance to models trained on human-annotated
data, both for sentence and word-level prediction.
| 2,021 |
Computation and Language
|
In-Order Chart-Based Constituent Parsing
|
We propose a novel in-order chart-based model for constituent parsing.
Compared with previous CKY-style and top-down models, our model gains
advantages from in-order traversal of a tree (rich features, lookahead
information and high efficiency) and makes a better use of structural knowledge
by encoding the history of decisions. Experiments on the Penn Treebank show
that our model outperforms previous chart-based models and achieves competitive
performance compared with other discriminative single models.
| 2,021 |
Computation and Language
|
VeeAlign: Multifaceted Context Representation using Dual Attention for
Ontology Alignment
|
Ontology Alignment is an important research problem applied to various fields
such as data integration, data transfer, data preparation, etc.
State-of-the-art (SOTA) Ontology Alignment systems typically use naive
domain-dependent approaches with handcrafted rules or domain-specific
architectures, making them unscalable and inefficient. In this work, we propose
VeeAlign, a Deep Learning based model that uses a novel dual-attention
mechanism to compute the contextualized representation of a concept which, in
turn, is used to discover alignments. By doing this, not only is our approach
able to exploit both syntactic and semantic information encoded in ontologies,
it is also, by design, flexible and scalable to different domains with minimal
effort. We evaluate our model on four different datasets from different domains
and languages, and establish its superiority through these results as well as
detailed ablation studies. The code and datasets used are available at
https://github.com/Remorax/VeeAlign.
| 2,021 |
Computation and Language
|
Effects of Layer Freezing on Transferring a Speech Recognition System to
Under-resourced Languages
|
In this paper, we investigate the effect of layer freezing on the
effectiveness of model transfer in the area of automatic speech recognition. We
experiment with Mozilla's DeepSpeech architecture on German and Swiss German
speech datasets and compare the results of either training from scratch vs.
transferring a pre-trained model. We compare different layer freezing schemes
and find that even freezing only one layer already significantly improves
results.
| 2,022 |
Computation and Language
|
Clinical Outcome Prediction from Admission Notes using Self-Supervised
Knowledge Integration
|
Outcome prediction from clinical text can prevent doctors from overlooking
possible risks and help hospitals to plan capacities. We simulate patients at
admission time, when decision support can be especially valuable, and
contribute a novel admission to discharge task with four common outcome
prediction targets: Diagnoses at discharge, procedures performed, in-hospital
mortality and length-of-stay prediction. The ideal system should infer outcomes
based on symptoms, pre-conditions and risk factors of a patient. We evaluate
the effectiveness of language models to handle this scenario and propose
clinical outcome pre-training to integrate knowledge about patient outcomes
from multiple public sources. We further present a simple method to incorporate
ICD code hierarchy into the models. We show that our approach improves
performance on the outcome tasks against several baselines. A detailed analysis
reveals further strengths of the model, including transferability, but also
weaknesses such as handling of vital values and inconsistencies in the
underlying data.
| 2,021 |
Computation and Language
|
Generate and Revise: Reinforcement Learning in Neural Poetry
|
Writers, poets, singers usually do not create their compositions in just one
breath. Text is revisited, adjusted, modified, rephrased, even multiple times,
in order to better convey meanings, emotions and feelings that the author wants
to express. Amongst the noble written arts, Poetry is probably the one that
needs to be elaborated the most, since the composition has to formally respect
predefined meter and rhyming schemes. In this paper, we propose a framework to
generate poems that are repeatedly revisited and corrected, as humans do, in
order to improve their overall quality. We frame the problem of revising poems
in the context of Reinforcement Learning and, in particular, using Proximal
Policy Optimization. Our model generates poems from scratch and it learns to
progressively adjust the generated text in order to match a target criterion.
We evaluate this approach in the case of matching a rhyming scheme, without
having any information on which words are responsible of creating rhymes and on
how to coherently alter the poem words. The proposed framework is general and,
with an appropriate reward shaping, it can be applied to other text generation
problems.
| 2,021 |
Computation and Language
|
Bias Out-of-the-Box: An Empirical Analysis of Intersectional
Occupational Biases in Popular Generative Language Models
|
The capabilities of natural language models trained on large-scale data have
increased immensely over the past few years. Open source libraries such as
HuggingFace have made these models easily available and accessible. While prior
research has identified biases in large language models, this paper considers
biases contained in the most popular versions of these models when applied
`out-of-the-box' for downstream tasks. We focus on generative language models
as they are well-suited for extracting biases inherited from training data.
Specifically, we conduct an in-depth analysis of GPT-2, which is the most
downloaded text generation model on HuggingFace, with over half a million
downloads per month. We assess biases related to occupational associations for
different protected categories by intersecting gender with religion, sexuality,
ethnicity, political affiliation, and continental name origin. Using a
template-based data collection pipeline, we collect 396K sentence completions
made by GPT-2 and find: (i) The machine-predicted jobs are less diverse and
more stereotypical for women than for men, especially for intersections; (ii)
Intersectional interactions are highly relevant for occupational associations,
which we quantify by fitting 262 logistic models; (iii) For most occupations,
GPT-2 reflects the skewed gender and ethnicity distribution found in US Labor
Bureau data, and even pulls the societally-skewed distribution towards gender
parity in cases where its predictions deviate from real labor market
observations. This raises the normative question of what language models should
learn - whether they should reflect or correct for existing inequalities.
| 2,021 |
Computation and Language
|
The Singleton Fallacy: Why Current Critiques of Language Models Miss the
Point
|
This paper discusses the current critique against neural network-based
Natural Language Understanding (NLU) solutions known as language models. We
argue that much of the current debate rests on an argumentation error that we
will refer to as the singleton fallacy: the assumption that language, meaning,
and understanding are single and uniform phenomena that are unobtainable by
(current) language models. By contrast, we will argue that there are many
different types of language use, meaning and understanding, and that (current)
language models are build with the explicit purpose of acquiring and
representing one type of structural understanding of language. We will argue
that such structural understanding may cover several different modalities, and
as such can handle several different types of meaning. Our position is that we
currently see no theoretical reason why such structural knowledge would be
insufficient to count as "real" understanding.
| 2,021 |
Computation and Language
|
Wake Word Detection with Streaming Transformers
|
Modern wake word detection systems usually rely on neural networks for
acoustic modeling. Transformers has recently shown superior performance over
LSTM and convolutional networks in various sequence modeling tasks with their
better temporal modeling power. However it is not clear whether this advantage
still holds for short-range temporal modeling like wake word detection.
Besides, the vanilla Transformer is not directly applicable to the task due to
its non-streaming nature and the quadratic time and space complexity. In this
paper we explore the performance of several variants of chunk-wise streaming
Transformers tailored for wake word detection in a recently proposed LF-MMI
system, including looking-ahead to the next chunk, gradient stopping, different
positional embedding methods and adding same-layer dependency between chunks.
Our experiments on the Mobvoi wake word dataset demonstrate that our proposed
Transformer model outperforms the baseline convolution network by 25% on
average in false rejection rate at the same false alarm rate with a comparable
model size, while still maintaining linear complexity w.r.t. the sequence
length.
| 2,021 |
Computation and Language
|
Unsupervised Abstractive Summarization of Bengali Text Documents
|
Abstractive summarization systems generally rely on large collections of
document-summary pairs. However, the performance of abstractive systems remains
a challenge due to the unavailability of parallel data for low-resource
languages like Bengali. To overcome this problem, we propose a graph-based
unsupervised abstractive summarization system in the single-document setting
for Bengali text documents, which requires only a Part-Of-Speech (POS) tagger
and a pre-trained language model trained on Bengali texts. We also provide a
human-annotated dataset with document-summary pairs to evaluate our abstractive
model and to support the comparison of future abstractive summarization systems
of the Bengali Language. We conduct experiments on this dataset and compare our
system with several well-established unsupervised extractive summarization
systems. Our unsupervised abstractive summarization model outperforms the
baselines without being exposed to any human-annotated reference summaries.
| 2,021 |
Computation and Language
|
A Hybrid Task-Oriented Dialog System with Domain and Task Adaptive
Pretraining
|
This paper describes our submission for the End-to-end Multi-domain Task
Completion Dialog shared task at the 9th Dialog System Technology Challenge
(DSTC-9). Participants in the shared task build an end-to-end task completion
dialog system which is evaluated by human evaluation and a user simulator based
automatic evaluation. Different from traditional pipelined approaches where
modules are optimized individually and suffer from cascading failure, we
propose an end-to-end dialog system that 1) uses Generative Pretraining 2
(GPT-2) as the backbone to jointly solve Natural Language Understanding, Dialog
State Tracking, and Natural Language Generation tasks, 2) adopts Domain and
Task Adaptive Pretraining to tailor GPT-2 to the dialog domain before
finetuning, 3) utilizes heuristic pre/post-processing rules that greatly
simplify the prediction tasks and improve generalizability, and 4) equips a
fault tolerance module to correct errors and inappropriate responses. Our
proposed method significantly outperforms baselines and ties for first place in
the official evaluation. We make our source code publicly available.
| 2,021 |
Computation and Language
|
A study of text representations in Hate Speech Detection
|
The pervasiveness of the Internet and social media have enabled the rapid and
anonymous spread of Hate Speech content on microblogging platforms such as
Twitter. Current EU and US legislation against hateful language, in conjunction
with the large amount of data produced in these platforms has led to automatic
tools being a necessary component of the Hate Speech detection task and
pipeline. In this study, we examine the performance of several, diverse text
representation techniques paired with multiple classification algorithms, on
the automatic Hate Speech detection and abusive language discrimination task.
We perform an experimental evaluation on binary and multiclass datasets, paired
with significance testing. Our results show that simple hate-keyword frequency
features (BoW) work best, followed by pre-trained word embeddings (GLoVe) as
well as N-gram graphs (NGGs): a graph-based representation which proved to
produce efficient, very low-dimensional but rich features for this task. A
combination of these representations paired with Logistic Regression or 3-layer
neural network classifiers achieved the best detection performance, in terms of
micro and macro F-measure.
| 2,021 |
Computation and Language
|
Joint Intent Detection and Slot Filling with Wheel-Graph Attention
Networks
|
Intent detection and slot filling are two fundamental tasks for building a
spoken language understanding (SLU) system. Multiple deep learning-based joint
models have demonstrated excellent results on the two tasks. In this paper, we
propose a new joint model with a wheel-graph attention network (Wheel-GAT)
which is able to model interrelated connections directly for intent detection
and slot filling. To construct a graph structure for utterances, we create
intent nodes, slot nodes, and directed edges. Intent nodes can provide
utterance-level semantic information for slot filling, while slot nodes can
also provide local keyword information for intent. Experiments show that our
model outperforms multiple baselines on two public datasets. Besides, we also
demonstrate that using Bidirectional Encoder Representation from Transformer
(BERT) model further boosts the performance in the SLU task.
| 2,021 |
Computation and Language
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.