Titles
stringlengths 6
220
| Abstracts
stringlengths 37
3.26k
| Years
int64 1.99k
2.02k
| Categories
stringclasses 1
value |
|---|---|---|---|
Language Acquisition is Embodied, Interactive, Emotive: a Research
Proposal
|
Humans' experience of the world is profoundly multimodal from the beginning,
so why do existing state-of-the-art language models only use text as a modality
to learn and represent semantic meaning? In this paper we review the literature
on the role of embodiment and emotion in the interactive setting of spoken
dialogue as necessary prerequisites for language learning for human children,
including how words in child vocabularies are largely concrete, then shift to
become more abstract as the children get older. We sketch a model of semantics
that leverages current transformer-based models and a word-level grounded
model, then explain the robot-dialogue system that will make use of our
semantic model, the setting for the system to learn language, and existing
benchmarks for evaluation.
| 2,021 |
Computation and Language
|
R2D2: Relational Text Decoding with Transformers
|
We propose a novel framework for modeling the interaction between graphical
structures and the natural language text associated with their nodes and edges.
Existing approaches typically fall into two categories. On group ignores the
relational structure by converting them into linear sequences and then utilize
the highly successful Seq2Seq models. The other side ignores the sequential
nature of the text by representing them as fixed-dimensional vectors and apply
graph neural networks. Both simplifications lead to information loss.
Our proposed method utilizes both the graphical structure as well as the
sequential nature of the texts. The input to our model is a set of text
segments associated with the nodes and edges of the graph, which are then
processed with a transformer encoder-decoder model, equipped with a
self-attention mechanism that is aware of the graphical relations between the
nodes containing the segments. This also allows us to use BERT-like models that
are already trained on large amounts of text.
While the proposed model has wide applications, we demonstrate its
capabilities on data-to-text generation tasks. Our approach compares favorably
against state-of-the-art methods in four tasks without tailoring the model
architecture. We also provide an early demonstration in a novel practical
application -- generating clinical notes from the medical entities mentioned
during clinical visits.
| 2,021 |
Computation and Language
|
GroupLink: An End-to-end Multitask Method for Word Grouping and Relation
Extraction in Form Understanding
|
Forms are a common type of document in real life and carry rich information
through textual contents and the organizational structure. To realize automatic
processing of forms, word grouping and relation extraction are two fundamental
and crucial steps after preliminary processing of optical character reader
(OCR). Word grouping is to aggregate words that belong to the same semantic
entity, and relation extraction is to predict the links between semantic
entities. Existing works treat them as two individual tasks, but these two
tasks are correlated and can reinforce each other. The grouping process will
refine the integrated representation of the corresponding entity, and the
linking process will give feedback to the grouping performance. For this
purpose, we acquire multimodal features from both textual data and layout
information and build an end-to-end model through multitask training to combine
word grouping and relation extraction to enhance performance on each task. We
validate our proposed method on a real-world, fully-annotated, noisy-scanned
benchmark, FUNSD, and extensive experiments demonstrate the effectiveness of
our method.
| 2,021 |
Computation and Language
|
What shall we do with an hour of data? Speech recognition for the un-
and under-served languages of Common Voice
|
This technical report describes the methods and results of a three-week
sprint to produce deployable speech recognition models for 31 under-served
languages of the Common Voice project. We outline the preprocessing steps,
hyperparameter selection, and resulting accuracy on official testing sets. In
addition to this we evaluate the models on multiple tasks: closed-vocabulary
speech recognition, pre-transcription, forced alignment, and key-word spotting.
The following experiments use Coqui STT, a toolkit for training and deployment
of neural Speech-to-Text models.
| 2,021 |
Computation and Language
|
Assessing the Syntactic Capabilities of Transformer-based Multilingual
Language Models
|
Multilingual Transformer-based language models, usually pretrained on more
than 100 languages, have been shown to achieve outstanding results in a wide
range of cross-lingual transfer tasks. However, it remains unknown whether the
optimization for different languages conditions the capacity of the models to
generalize over syntactic structures, and how languages with syntactic
phenomena of different complexity are affected. In this work, we explore the
syntactic generalization capabilities of the monolingual and multilingual
versions of BERT and RoBERTa. More specifically, we evaluate the syntactic
generalization potential of the models on English and Spanish tests, comparing
the syntactic abilities of monolingual and multilingual models on the same
language (English), and of multilingual models on two different languages
(English and Spanish). For English, we use the available SyntaxGym test suite;
for Spanish, we introduce SyntaxGymES, a novel ensemble of targeted syntactic
tests in Spanish, designed to evaluate the syntactic generalization
capabilities of language models through the SyntaxGym online platform.
| 2,021 |
Computation and Language
|
Word-level Human Interpretable Scoring Mechanism for Novel Text
Detection Using Tsetlin Machines
|
Recent research in novelty detection focuses mainly on document-level
classification, employing deep neural networks (DNN). However, the black-box
nature of DNNs makes it difficult to extract an exact explanation of why a
document is considered novel. In addition, dealing with novelty at the
word-level is crucial to provide a more fine-grained analysis than what is
available at the document level. In this work, we propose a Tsetlin machine
(TM)-based architecture for scoring individual words according to their
contribution to novelty. Our approach encodes a description of the novel
documents using the linguistic patterns captured by TM clauses. We then adopt
this description to measure how much a word contributes to making documents
novel. Our experimental results demonstrate how our approach breaks down
novelty into interpretable phrases, successfully measuring novelty.
| 2,021 |
Computation and Language
|
Speech2Slot: An End-to-End Knowledge-based Slot Filling from Speech
|
In contrast to conventional pipeline Spoken Language Understanding (SLU)
which consists of automatic speech recognition (ASR) and natural language
understanding (NLU), end-to-end SLU infers the semantic meaning directly from
speech and overcomes the error propagation caused by ASR. End-to-end slot
filling (SF) from speech is an essential component of end-to-end SLU, and is
usually regarded as a sequence-to-sequence generation problem, heavily relied
on the performance of language model of ASR. However, it is hard to generate a
correct slot when the slot is out-of-vovabulary (OOV) in training data,
especially when a slot is an anti-linguistic entity without grammatical rule.
Inspired by object detection in computer vision that is to detect the object
from an image, we consider SF as the task of slot detection from speech. In
this paper, we formulate the SF task as a matching task and propose an
end-to-end knowledge-based SF model, named Speech-to-Slot (Speech2Slot), to
leverage knowledge to detect the boundary of a slot from the speech. We also
release a large-scale dataset of Chinese speech for slot filling, containing
more than 830,000 samples. The experiments show that our approach is markedly
superior to the conventional pipeline SLU approach, and outperforms the
state-of-the-art end-to-end SF approach with 12.51% accuracy improvement.
| 2,021 |
Computation and Language
|
EL-Attention: Memory Efficient Lossless Attention for Generation
|
Transformer model with multi-head attention requires caching intermediate
results for efficient inference in generation tasks. However, cache brings new
memory-related costs and prevents leveraging larger batch size for faster
speed. We propose memory-efficient lossless attention (called EL-attention) to
address this issue. It avoids heavy operations for building multi-head keys and
values, cache for them is not needed. EL-attention constructs an ensemble of
attention results by expanding query while keeping key and value shared. It
produces the same result as multi-head attention with less GPU memory and
faster inference speed. We conduct extensive experiments on Transformer, BART,
and GPT-2 for summarization and question generation tasks. The results show
EL-attention speeds up existing models by 1.6x to 5.3x without accuracy loss.
| 2,021 |
Computation and Language
|
Rationalization through Concepts
|
Automated predictions require explanations to be interpretable by humans. One
type of explanation is a rationale, i.e., a selection of input features such as
relevant text snippets from which the model computes the outcome. However, a
single overall selection does not provide a complete explanation, e.g.,
weighing several aspects for decisions. To this end, we present a novel
self-interpretable model called ConRAT. Inspired by how human explanations for
high-level decisions are often based on key concepts, ConRAT extracts a set of
text snippets as concepts and infers which ones are described in the document.
Then, it explains the outcome with a linear aggregation of concepts. Two
regularizers drive ConRAT to build interpretable concepts. In addition, we
propose two techniques to boost the rationale and predictive performance
further. Experiments on both single- and multi-aspect sentiment classification
tasks show that ConRAT is the first to generate concepts that align with human
rationalization while using only the overall label. Further, it outperforms
state-of-the-art methods trained on each aspect label independently.
| 2,021 |
Computation and Language
|
Investigating the Reordering Capability in CTC-based Non-Autoregressive
End-to-End Speech Translation
|
We study the possibilities of building a non-autoregressive speech-to-text
translation model using connectionist temporal classification (CTC), and use
CTC-based automatic speech recognition as an auxiliary task to improve the
performance. CTC's success on translation is counter-intuitive due to its
monotonicity assumption, so we analyze its reordering capability. Kendall's tau
distance is introduced as the quantitative metric, and gradient-based
visualization provides an intuitive way to take a closer look into the model.
Our analysis shows that transformer encoders have the ability to change the
word order and points out the future research direction that worth being
explored more on non-autoregressive speech translation.
| 2,021 |
Computation and Language
|
Can You Traducir This? Machine Translation for Code-Switched Input
|
Code-Switching (CSW) is a common phenomenon that occurs in multilingual
geographic or social contexts, which raises challenging problems for natural
language processing tools. We focus here on Machine Translation (MT) of CSW
texts, where we aim to simultaneously disentangle and translate the two mixed
languages. Due to the lack of actual translated CSW data, we generate
artificial training data from regular parallel texts. Experiments show this
training strategy yields MT systems that surpass multilingual systems for
code-switched texts. These results are confirmed in an alternative task aimed
at providing contextual translations for a L2 writing assistant.
| 2,021 |
Computation and Language
|
Benchmarking down-scaled (not so large) pre-trained language models
|
Large Transformer-based language models are pre-trained on corpora of varying
sizes, for a different number of steps and with different batch sizes. At the
same time, more fundamental components, such as the pre-training objective or
architectural hyperparameters, are modified. In total, it is therefore
difficult to ascribe changes in performance to specific factors. Since
searching the hyperparameter space over the full systems is too costly, we
pre-train down-scaled versions of several popular Transformer-based
architectures on a common pre-training corpus and benchmark them on a subset of
the GLUE tasks (Wang et al., 2018). Specifically, we systematically compare
three pre-training objectives for different shape parameters and model sizes,
while also varying the number of pre-training steps and the batch size. In our
experiments MLM + NSP (BERT-style) consistently outperforms MLM (RoBERTa-style)
as well as the standard LM objective. Furthermore, we find that additional
compute should be mainly allocated to an increased model size, while training
for more steps is inefficient. Based on these observations, as a final step we
attempt to scale up several systems using compound scaling (Tan and Le, 2019)
adapted to Transformer-based language models.
| 2,021 |
Computation and Language
|
Conversational Entity Linking: Problem Definition and Datasets
|
Machine understanding of user utterances in conversational systems is of
utmost importance for enabling engaging and meaningful conversations with
users. Entity Linking (EL) is one of the means of text understanding, with
proven efficacy for various downstream tasks in information retrieval. In this
paper, we study entity linking for conversational systems. To develop a better
understanding of what EL in a conversational setting entails, we analyze a
large number of dialogues from existing conversational datasets and annotate
references to concepts, named entities, and personal entities using
crowdsourcing. Based on the annotated dialogues, we identify the main
characteristics of conversational entity linking. Further, we report on the
performance of traditional EL systems on our Conversational Entity Linking
dataset, ConEL, and present an extension to these methods to better fit the
conversational setting. The resources released with this paper include
annotated datasets, detailed descriptions of crowdsourcing setups, as well as
the annotations produced by various EL systems. These new resources allow for
an investigation of how the role of entities in conversations is different from
that in documents or isolated short text utterances like queries and tweets,
and complement existing conversational datasets.
| 2,021 |
Computation and Language
|
Role of Artificial Intelligence in Detection of Hateful Speech for
Hinglish Data on Social Media
|
Social networking platforms provide a conduit to disseminate our ideas, views
and thoughts and proliferate information. This has led to the amalgamation of
English with natively spoken languages. Prevalence of Hindi-English code-mixed
data (Hinglish) is on the rise with most of the urban population all over the
world. Hate speech detection algorithms deployed by most social networking
platforms are unable to filter out offensive and abusive content posted in
these code-mixed languages. Thus, the worldwide hate speech detection rate of
around 44% drops even more considering the content in Indian colloquial
languages and slangs. In this paper, we propose a methodology for efficient
detection of unstructured code-mix Hinglish language. Fine-tuning based
approaches for Hindi-English code-mixed language are employed by utilizing
contextual based embeddings such as ELMo (Embeddings for Language Models),
FLAIR, and transformer-based BERT (Bidirectional Encoder Representations from
Transformers). Our proposed approach is compared against the pre-existing
methods and results are compared for various datasets. Our model outperforms
the other methods and frameworks.
| 2,021 |
Computation and Language
|
BERT is to NLP what AlexNet is to CV: Can Pre-Trained Language Models
Identify Analogies?
|
Analogies play a central role in human commonsense reasoning. The ability to
recognize analogies such as "eye is to seeing what ear is to hearing",
sometimes referred to as analogical proportions, shape how we structure
knowledge and understand language. Surprisingly, however, the task of
identifying such analogies has not yet received much attention in the language
model era. In this paper, we analyze the capabilities of transformer-based
language models on this unsupervised task, using benchmarks obtained from
educational settings, as well as more commonly used datasets. We find that
off-the-shelf language models can identify analogies to a certain extent, but
struggle with abstract and complex relations, and results are highly sensitive
to model architecture and hyperparameters. Overall the best results were
obtained with GPT-2 and RoBERTa, while configurations using BERT were not able
to outperform word embedding models. Our results raise important questions for
future work about how, and to what extent, pre-trained language models capture
knowledge about abstract semantic relations.
| 2,022 |
Computation and Language
|
Backretrieval: An Image-Pivoted Evaluation Metric for Cross-Lingual Text
Representations Without Parallel Corpora
|
Cross-lingual text representations have gained popularity lately and act as
the backbone of many tasks such as unsupervised machine translation and
cross-lingual information retrieval, to name a few. However, evaluation of such
representations is difficult in the domains beyond standard benchmarks due to
the necessity of obtaining domain-specific parallel language data across
different pairs of languages. In this paper, we propose an automatic metric for
evaluating the quality of cross-lingual textual representations using images as
a proxy in a paired image-text evaluation dataset. Experimentally,
Backretrieval is shown to highly correlate with ground truth metrics on
annotated datasets, and our analysis shows statistically significant
improvements over baselines. Our experiments conclude with a case study on a
recipe dataset without parallel cross-lingual data. We illustrate how to judge
cross-lingual embedding quality with Backretrieval, and validate the outcome
with a small human study.
| 2,021 |
Computation and Language
|
Designing an Automatic Agent for Repeated Language based Persuasion
Games
|
Persuasion games are fundamental in economics and AI research and serve as
the basis for important applications. However, work on this setup assumes
communication with stylized messages that do not consist of rich human
language. In this paper we consider a repeated sender (expert) -- receiver
(decision maker) game, where the sender is fully informed about the state of
the world and aims to persuade the receiver to accept a deal by sending one of
several possible natural language reviews. We design an automatic expert that
plays this repeated game, aiming to achieve the maximal payoff. Our expert is
implemented within the Monte Carlo Tree Search (MCTS) algorithm, with deep
learning models that exploit behavioral and linguistic signals in order to
predict the next action of the decision maker, and the future payoff of the
expert given the state of the game and a candidate review. We demonstrate the
superiority of our expert over strong baselines, its adaptability to different
decision makers, and that its selected reviews are nicely adapted to the
proposed deal.
| 2,022 |
Computation and Language
|
Towards transparency in NLP shared tasks
|
This article reports on a survey carried out across the Natural Language
Processing (NLP) community. The survey aimed to capture the opinions of the
research community on issues surrounding shared tasks, with respect to both
participation and organisation. Amongst the 175 responses received, both
positive and negative observations were made. We carried out and report on an
extensive analysis of these responses, which leads us to propose a Shared Task
Organisation Checklist that could support future participants and organisers.
The proposed Checklist is flexible enough to accommodate the wide diversity of
shared tasks in our field and its goal is not to be prescriptive, but rather to
serve as a tool that encourages shared task organisers to foreground ethical
behaviour, beginning with the common issues that the 175 respondents deemed
important. Its usage would not only serve as an instrument to reflect on
important aspects of shared tasks, but would also promote increased
transparency around them.
| 2,021 |
Computation and Language
|
Joint Text and Label Generation for Spoken Language Understanding
|
Generalization is a central problem in machine learning, especially when data
is limited. Using prior information to enforce constraints is the principled
way of encouraging generalization. In this work, we propose to leverage the
prior information embedded in pretrained language models (LM) to improve
generalization for intent classification and slot labeling tasks with limited
training data. Specifically, we extract prior knowledge from pretrained LM in
the form of synthetic data, which encode the prior implicitly. We fine-tune the
LM to generate an augmented language, which contains not only text but also
encodes both intent labels and slot labels. The generated synthetic data can be
used to train a classifier later. Since the generated data may contain noise,
we rephrase the learning from generated data as learning with noisy labels. We
then utilize the mixout regularization for the classifier and prove its
effectiveness to resist label noise in generated data. Empirically, our method
demonstrates superior performance and outperforms the baseline by a large
margin.
| 2,021 |
Computation and Language
|
Towards Using Diachronic Distributed Word Representations as Models of
Lexical Development
|
Recent work has shown that distributed word representations can encode
abstract information from child-directed speech. In this paper, we use
diachronic distributed word representations to perform temporal modeling and
analysis of lexical development in children. Unlike all previous work, we use
temporally sliced corpus to learn distributed word representations of
child-speech and child-directed speech under a curriculum-learning setting. In
our experiments, we perform a lexical categorization task to plot the semantic
and syntactic knowledge acquisition trajectories in children. Next, we perform
linear mixed-effects modeling over the diachronic representational changes to
study the role of input word frequencies in the rate of word acquisition in
children. We also perform a fine-grained analysis of lexical knowledge transfer
from adults to children using Representational Similarity Analysis. Finally, we
perform a qualitative analysis of the diachronic representations from our
model, which reveals the grounding and word associations in the mental lexicon
of children. Our experiments demonstrate the ease of usage and effectiveness of
diachronic distributed word representations in modeling lexical development.
| 2,021 |
Computation and Language
|
Integrating extracted information from bert and multiple embedding
methods with the deep neural network for humour detection
|
Humour detection from sentences has been an interesting and challenging task
in the last few years. In attempts to highlight humour detection, most research
was conducted using traditional approaches of embedding, e.g., Word2Vec or
Glove. Recently BERT sentence embedding has also been used for this task. In
this paper, we propose a framework for humour detection in short texts taken
from news headlines. Our proposed framework (IBEN) attempts to extract
information from written text via the use of different layers of BERT. After
several trials, weights were assigned to different layers of the BERT model.
The extracted information was then sent to a Bi-GRU neural network as an
embedding matrix. We utilized the properties of some external embedding models.
A multi-kernel convolution in our neural network was also employed to extract
higher-level sentence representations. This framework performed very well on
the task of humour detection.
| 2,021 |
Computation and Language
|
kdehumor at semeval-2020 task 7: a neural network model for detecting
funniness in dataset humicroedit
|
This paper describes our contribution to SemEval-2020 Task 7: Assessing Humor
in Edited News Headlines. Here we present a method based on a deep neural
network. In recent years, quite some attention has been devoted to humor
production and perception. Our team KdeHumor employs recurrent neural network
models including Bi-Directional LSTMs (BiLSTMs). Moreover, we utilize the
state-of-the-art pre-trained sentence embedding techniques. We analyze the
performance of our method and demonstrate the contribution of each component of
our architecture.
| 2,021 |
Computation and Language
|
Restoring Hebrew Diacritics Without a Dictionary
|
We demonstrate that it is feasible to diacritize Hebrew script without any
human-curated resources other than plain diacritized text. We present NAKDIMON,
a two-layer character level LSTM, that performs on par with much more
complicated curation-dependent systems, across a diverse array of modern Hebrew
sources.
| 2,022 |
Computation and Language
|
Including Signed Languages in Natural Language Processing
|
Signed languages are the primary means of communication for many deaf and
hard of hearing individuals. Since signed languages exhibit all the fundamental
linguistic properties of natural language, we believe that tools and theories
of Natural Language Processing (NLP) are crucial towards its modeling. However,
existing research in Sign Language Processing (SLP) seldom attempt to explore
and leverage the linguistic organization of signed languages. This position
paper calls on the NLP community to include signed languages as a research area
with high social and scientific impact. We first discuss the linguistic
properties of signed languages to consider during their modeling. Then, we
review the limitations of current SLP models and identify the open challenges
to extend NLP to signed languages. Finally, we urge (1) the adoption of an
efficient tokenization method; (2) the development of linguistically-informed
models; (3) the collection of real-world signed language data; (4) the
inclusion of local signed language communities as an active and leading voice
in the direction of research.
| 2,021 |
Computation and Language
|
Doing Natural Language Processing in A Natural Way: An NLP toolkit based
on object-oriented knowledge base and multi-level grammar base
|
We introduce an NLP toolkit based on object-oriented knowledge base and
multi-level grammar base. This toolkit focuses on semantic parsing, it also has
abilities to discover new knowledge and grammar automatically, new discovered
knowledge and grammar will be identified by human, and will be used to update
the knowledge base and grammar base. This process can be iterated many times to
improve the toolkit continuously.
| 2,021 |
Computation and Language
|
Addressing "Documentation Debt" in Machine Learning Research: A
Retrospective Datasheet for BookCorpus
|
Recent literature has underscored the importance of dataset documentation
work for machine learning, and part of this work involves addressing
"documentation debt" for datasets that have been used widely but documented
sparsely. This paper aims to help address documentation debt for BookCorpus, a
popular text dataset for training large language models. Notably, researchers
have used BookCorpus to train OpenAI's GPT-N models and Google's BERT models,
even though little to no documentation exists about the dataset's motivation,
composition, collection process, etc. We offer a preliminary datasheet that
provides key context and information about BookCorpus, highlighting several
notable deficiencies. In particular, we find evidence that (1) BookCorpus
likely violates copyright restrictions for many books, (2) BookCorpus contains
thousands of duplicated books, and (3) BookCorpus exhibits significant skews in
genre representation. We also find hints of other potential deficiencies that
call for future research, including problematic content, potential skews in
religious representation, and lopsided author contributions. While more work
remains, this initial effort to provide a datasheet for BookCorpus adds to
growing literature that urges more careful and systematic documentation for
machine learning datasets.
| 2,021 |
Computation and Language
|
The Summary Loop: Learning to Write Abstractive Summaries Without
Examples
|
This work presents a new approach to unsupervised abstractive summarization
based on maximizing a combination of coverage and fluency for a given length
constraint. It introduces a novel method that encourages the inclusion of key
terms from the original document into the summary: key terms are masked out of
the original document and must be filled in by a coverage model using the
current generated summary. A novel unsupervised training procedure leverages
this coverage model along with a fluency model to generate and score summaries.
When tested on popular news summarization datasets, the method outperforms
previous unsupervised methods by more than 2 R-1 points, and approaches results
of competitive supervised methods. Our model attains higher levels of
abstraction with copied passages roughly two times shorter than prior work, and
learns to compress and merge sentences without supervision.
| 2,020 |
Computation and Language
|
News Headline Grouping as a Challenging NLU Task
|
Recent progress in Natural Language Understanding (NLU) has seen the latest
models outperform human performance on many standard tasks. These impressive
results have led the community to introspect on dataset limitations, and
iterate on more nuanced challenges. In this paper, we introduce the task of
HeadLine Grouping (HLG) and a corresponding dataset (HLGD) consisting of 20,056
pairs of news headlines, each labeled with a binary judgement as to whether the
pair belongs within the same group. On HLGD, human annotators achieve high
performance of around 0.9 F-1, while current state-of-the art Transformer
models only reach 0.75 F-1, opening the path for further improvements. We
further propose a novel unsupervised Headline Generator Swap model for the task
of HeadLine Grouping that achieves within 3 F-1 of the best supervised model.
Finally, we analyze high-performing models with consistency tests, and find
that models are not consistent in their predictions, revealing modeling limits
of current architectures.
| 2,021 |
Computation and Language
|
What's The Latest? A Question-driven News Chatbot
|
This work describes an automatic news chatbot that draws content from a
diverse set of news articles and creates conversations with a user about the
news. Key components of the system include the automatic organization of news
articles into topical chatrooms, integration of automatically generated
questions into the conversation, and a novel method for choosing which
questions to present which avoids repetitive suggestions. We describe the
algorithmic framework and present the results of a usability study that shows
that news readers using the system successfully engage in multi-turn
conversations about specific news stories.
| 2,020 |
Computation and Language
|
Could you give me a hint? Generating inference graphs for defeasible
reasoning
|
Defeasible reasoning is the mode of reasoning where conclusions can be
overturned by taking into account new evidence. A commonly used method in
cognitive science and logic literature is to handcraft argumentation supporting
inference graphs. While humans find inference graphs very useful for reasoning,
constructing them at scale is difficult. In this paper, we automatically
generate such inference graphs through transfer learning from another NLP task
that shares the kind of reasoning that inference graphs support. Through
automated metrics and human evaluation, we find that our method generates
meaningful graphs for the defeasible inference task. Human accuracy on this
task improves by 20% by consulting the generated graphs. Our findings open up
exciting new research avenues for cases where machine reasoning can help human
reasoning. (A dataset of 230,000 influence graphs for each defeasible query is
located at: https://tinyurl.com/defeasiblegraphs.)
| 2,021 |
Computation and Language
|
UIUC_BioNLP at SemEval-2021 Task 11: A Cascade of Neural Models for
Structuring Scholarly NLP Contributions
|
We propose a cascade of neural models that performs sentence classification,
phrase recognition, and triple extraction to automatically structure the
scholarly contributions of NLP publications. To identify the most important
contribution sentences in a paper, we used a BERT-based classifier with
positional features (Subtask 1). A BERT-CRF model was used to recognize and
characterize relevant phrases in contribution sentences (Subtask 2). We
categorized the triples into several types based on whether and how their
elements were expressed in text, and addressed each type using separate
BERT-based classifiers as well as rules (Subtask 3). Our system was officially
ranked second in Phase 1 evaluation and first in both parts of Phase 2
evaluation. After fixing a submission error in Pharse 1, our approach yields
the best results overall. In this paper, in addition to a system description,
we also provide further analysis of our results, highlighting its strengths and
limitations. We make our code publicly available at
https://github.com/Liu-Hy/nlp-contrib-graph.
| 2,021 |
Computation and Language
|
Incorporating Commonsense Knowledge Graph in Pretrained Models for
Social Commonsense Tasks
|
Pretrained language models have excelled at many NLP tasks recently; however,
their social intelligence is still unsatisfactory. To enable this, machines
need to have a more general understanding of our complicated world and develop
the ability to perform commonsense reasoning besides fitting the specific
downstream tasks. External commonsense knowledge graphs (KGs), such as
ConceptNet, provide rich information about words and their relationships. Thus,
towards general commonsense learning, we propose two approaches to
\emph{implicitly} and \emph{explicitly} infuse such KGs into pretrained
language models. We demonstrate our proposed methods perform well on SocialIQA,
a social commonsense reasoning task, in both limited and full training data
regimes.
| 2,021 |
Computation and Language
|
Improving Lexically Constrained Neural Machine Translation with
Source-Conditioned Masked Span Prediction
|
Accurate terminology translation is crucial for ensuring the practicality and
reliability of neural machine translation (NMT) systems. To address this,
lexically constrained NMT explores various methods to ensure pre-specified
words and phrases appear in the translation output. However, in many cases,
those methods are studied on general domain corpora, where the terms are mostly
uni- and bi-grams (>98%). In this paper, we instead tackle a more challenging
setup consisting of domain-specific corpora with much longer n-gram and highly
specialized terms. Inspired by the recent success of masked span prediction
models, we propose a simple and effective training strategy that achieves
consistent improvements on both terminology and sentence-level translation for
three domain-specific corpora in two language pairs.
| 2,021 |
Computation and Language
|
Probabilistic modeling of rational communication with conditionals
|
While a large body of work has scrutinized the meaning of conditional
sentences, considerably less attention has been paid to formal models of their
pragmatic use and interpretation. Here, we take a probabilistic approach to
pragmatic reasoning about indicative conditionals which flexibly integrates
gradient beliefs about richly structured world states. We model listeners'
update of their prior beliefs about the causal structure of the world and the
joint probabilities of the consequent and antecedent based on assumptions about
the speaker's utterance production protocol. We show that, when supplied with
natural contextual assumptions, our model uniformly explains a number of
inferences attested in the literature, including epistemic inferences,
conditional perfection and the dependency between antecedent and consequent of
a conditional. We argue that this approach also helps explain three puzzles
introduced by Douven (2012) about updating with conditionals: depending on the
utterance context, the listener's belief in the antecedent may increase,
decrease or remain unchanged.
| 2,022 |
Computation and Language
|
OCHADAI-KYOTO at SemEval-2021 Task 1: Enhancing Model Generalization and
Robustness for Lexical Complexity Prediction
|
We propose an ensemble model for predicting the lexical complexity of words
and multiword expressions (MWEs). The model receives as input a sentence with a
target word or MWEand outputs its complexity score. Given that a key challenge
with this task is the limited size of annotated data, our model relies on
pretrained contextual representations from different state-of-the-art
transformer-based language models (i.e., BERT and RoBERTa), and on a variety of
training methods for further enhancing model generalization and
robustness:multi-step fine-tuning and multi-task learning, and adversarial
training. Additionally, we propose to enrich contextual representations by
adding hand-crafted features during training. Our model achieved competitive
results and ranked among the top-10 systems in both sub-tasks.
| 2,021 |
Computation and Language
|
Evaluating Gender Bias in Natural Language Inference
|
Gender-bias stereotypes have recently raised significant ethical concerns in
natural language processing. However, progress in detection and evaluation of
gender bias in natural language understanding through inference is limited and
requires further investigation. In this work, we propose an evaluation
methodology to measure these biases by constructing a challenge task that
involves pairing gender-neutral premises against a gender-specific hypothesis.
We use our challenge task to investigate state-of-the-art NLI models on the
presence of gender stereotypes using occupations. Our findings suggest that
three models (BERT, RoBERTa, BART) trained on MNLI and SNLI datasets are
significantly prone to gender-induced prediction errors. We also find that
debiasing techniques such as augmenting the training dataset to ensure a
gender-balanced dataset can help reduce such bias in certain cases.
| 2,021 |
Computation and Language
|
!Qu\'e maravilla! Multimodal Sarcasm Detection in Spanish: a Dataset and
a Baseline
|
We construct the first ever multimodal sarcasm dataset for Spanish. The
audiovisual dataset consists of sarcasm annotated text that is aligned with
video and audio. The dataset represents two varieties of Spanish, a Latin
American variety and a Peninsular Spanish variety, which ensures a wider
dialectal coverage for this global language. We present several models for
sarcasm detection that will serve as baselines in the future research. Our
results show that results with text only (89%) are worse than when combining
text with audio (91.9%). Finally, the best results are obtained when combining
all the modalities: text, audio and video (93.1%).
| 2,021 |
Computation and Language
|
Supporting Land Reuse of Former Open Pit Mining Sites using Text
Classification and Active Learning
|
Open pit mines left many regions worldwide inhospitable or uninhabitable. To
put these regions back into use, entire stretches of land must be
renaturalized. For the sustainable subsequent use or transfer to a new primary
use, many contaminated sites and soil information have to be permanently
managed. In most cases, this information is available in the form of expert
reports in unstructured data collections or file folders, which in the best
case are digitized. Due to size and complexity of the data, it is difficult for
a single person to have an overview of this data in order to be able to make
reliable statements. This is one of the most important obstacles to the rapid
transfer of these areas to after-use. An information-based approach to this
issue supports fulfilling several Sustainable Development Goals regarding
environment issues, health and climate action. We use a stack of Optical
Character Recognition, Text Classification, Active Learning and Geographic
Information System Visualization to effectively mine and visualize this
information. Subsequently, we link the extracted information to geographic
coordinates and visualize them using a Geographic Information System. Active
Learning plays a vital role because our dataset provides no training data. In
total, we process nine categories and actively learn their representation in
our dataset. We evaluate the OCR, Active Learning and Text Classification
separately to report the performance of the system. Active Learning and text
classification results are twofold: Whereas our categories about restrictions
work sufficient ($>$.85 F1), the seven topic-oriented categories were
complicated for human coders and hence the results achieved mediocre evaluation
scores ($<$.70 F1).
| 2,021 |
Computation and Language
|
Discrete representations in neural models of spoken language
|
The distributed and continuous representations used by neural networks are at
odds with representations employed in linguistics, which are typically
symbolic. Vector quantization has been proposed as a way to induce discrete
neural representations that are closer in nature to their linguistic
counterparts. However, it is not clear which metrics are the best-suited to
analyze such discrete representations. We compare the merits of four commonly
used metrics in the context of weakly supervised models of spoken language. We
compare the results they show when applied to two different models, while
systematically studying the effect of the placement and size of the
discretization layer. We find that different evaluation regimes can give
inconsistent results. While we can attribute them to the properties of the
different metrics in most cases, one point of concern remains: the use of
minimal pairs of phoneme triples as stimuli disadvantages larger discrete unit
inventories, unlike metrics applied to complete utterances. Furthermore, while
in general vector quantization induces representations that correlate with
units posited in linguistics, the strength of this correlation is only
moderate.
| 2,021 |
Computation and Language
|
Unsupervised Knowledge Graph Alignment by Probabilistic Reasoning and
Semantic Embedding
|
Knowledge Graph (KG) alignment is to discover the mappings (i.e., equivalent
entities, relations, and others) between two KGs. The existing methods can be
divided into the embedding-based models, and the conventional reasoning and
lexical matching based systems. The former compute the similarity of entities
via their cross-KG embeddings, but they usually rely on an ideal supervised
learning setting for good performance and lack appropriate reasoning to avoid
logically wrong mappings; while the latter address the reasoning issue but are
poor at utilizing the KG graph structures and the entity contexts. In this
study, we aim at combining the above two solutions and thus propose an
iterative framework named PRASE which is based on probabilistic reasoning and
semantic embedding. It learns the KG embeddings via entity mappings from a
probabilistic reasoning system named PARIS, and feeds the resultant entity
mappings and embeddings back into PARIS for augmentation. The PRASE framework
is compatible with different embedding-based models, and our experiments on
multiple datasets have demonstrated its state-of-the-art performance.
| 2,021 |
Computation and Language
|
OutFlip: Generating Out-of-Domain Samples for Unknown Intent Detection
with Natural Language Attack
|
Out-of-domain (OOD) input detection is vital in a task-oriented dialogue
system since the acceptance of unsupported inputs could lead to an incorrect
response of the system. This paper proposes OutFlip, a method to generate
out-of-domain samples using only in-domain training dataset automatically. A
white-box natural language attack method HotFlip is revised to generate
out-of-domain samples instead of adversarial examples. Our evaluation results
showed that integrating OutFlip-generated out-of-domain samples into the
training dataset could significantly improve an intent classification model's
out-of-domain detection performance.
| 2,021 |
Computation and Language
|
Priberam Labs at the NTCIR-15 SHINRA2020-ML: Classification Task
|
Wikipedia is an online encyclopedia available in 285 languages. It composes
an extremely relevant Knowledge Base (KB), which could be leveraged by
automatic systems for several purposes. However, the structure and organisation
of such information are not prone to automatic parsing and understanding and it
is, therefore, necessary to structure this knowledge. The goal of the current
SHINRA2020-ML task is to leverage Wikipedia pages in order to categorise their
corresponding entities across 268 hierarchical categories, belonging to the
Extended Named Entity (ENE) ontology. In this work, we propose three distinct
models based on the contextualised embeddings yielded by Multilingual BERT. We
explore the performances of a linear layer with and without explicit usage of
the ontology's hierarchy, and a Gated Recurrent Units (GRU) layer. We also test
several pooling strategies to leverage BERT's embeddings and selection criteria
based on the labels' scores. We were able to achieve good performance across a
large variety of languages, including those not seen during the fine-tuning
process (zero-shot languages).
| 2,021 |
Computation and Language
|
Priberam at MESINESP Multi-label Classification of Medical Texts Task
|
Medical articles provide current state of the art treatments and diagnostics
to many medical practitioners and professionals. Existing public databases such
as MEDLINE contain over 27 million articles, making it difficult to extract
relevant content without the use of efficient search engines. Information
retrieval tools are crucial in order to navigate and provide meaningful
recommendations for articles and treatments. Classifying these articles into
broader medical topics can improve the retrieval of related articles. The set
of medical labels considered for the MESINESP task is on the order of several
thousands of labels (DeCS codes), which falls under the extreme multi-label
classification problem. The heterogeneous and highly hierarchical structure of
medical topics makes the task of manually classifying articles extremely
laborious and costly. It is, therefore, crucial to automate the process of
classification. Typical machine learning algorithms become computationally
demanding with such a large number of labels and achieving better recall on
such datasets becomes an unsolved problem.
This work presents Priberam's participation at the BioASQ task Mesinesp. We
address the large multi-label classification problem through the use of four
different models: a Support Vector Machine (SVM), a customised search engine
(Priberam Search), a BERT based classifier, and a SVM-rank ensemble of all the
previous models. Results demonstrate that all three individual models perform
well and the best performance is achieved by their ensemble, granting Priberam
the 6th place in the present challenge and making it the 2nd best team.
| 2,021 |
Computation and Language
|
NLP for Climate Policy: Creating a Knowledge Platform for Holistic and
Effective Climate Action
|
Climate change is a burning issue of our time, with the Sustainable
Development Goal (SDG) 13 of the United Nations demanding global climate
action. Realizing the urgency, in 2015 in Paris, world leaders signed an
agreement committing to taking voluntary action to reduce carbon emissions.
However, the scale, magnitude, and climate action processes vary globally,
especially between developed and developing countries. Therefore, from
parliament to social media, the debates and discussions on climate change
gather data from wide-ranging sources essential to the policy design and
implementation. The downside is that we do not currently have the mechanisms to
pool the worldwide dispersed knowledge emerging from the structured and
unstructured data sources.
The paper thematically discusses how NLP techniques could be employed in
climate policy research and contribute to society's good at large. In
particular, we exemplify symbiosis of NLP and Climate Policy Research via four
methodologies. The first one deals with the major topics related to climate
policy using automated content analysis. We investigate the opinions
(sentiments) of major actors' narratives towards climate policy in the second
methodology. The third technique explores the climate actors' beliefs towards
pro or anti-climate orientation. Finally, we discuss developing a Climate
Knowledge Graph.
The present theme paper further argues that creating a knowledge platform
would help in the formulation of a holistic climate policy and effective
climate action. Such a knowledge platform would integrate the policy actors'
varied opinions from different social sectors like government, business, civil
society, and the scientific community. The research outcome will add value to
effective climate action because policymakers can make informed decisions by
looking at the diverse public opinion on a comprehensive platform.
| 2,021 |
Computation and Language
|
How Reliable are Model Diagnostics?
|
In the pursuit of a deeper understanding of a model's behaviour, there is
recent impetus for developing suites of probes aimed at diagnosing models
beyond simple metrics like accuracy or BLEU. This paper takes a step back and
asks an important and timely question: how reliable are these diagnostics in
providing insight into models and training setups? We critically examine three
recent diagnostic tests for pre-trained language models, and find that
likelihood-based and representation-based model diagnostics are not yet as
reliable as previously assumed. Based on our empirical findings, we also
formulate recommendations for practitioners and researchers.
| 2,022 |
Computation and Language
|
BertGCN: Transductive Text Classification by Combining GCN and BERT
|
In this work, we propose BertGCN, a model that combines large scale
pretraining and transductive learning for text classification. BertGCN
constructs a heterogeneous graph over the dataset and represents documents as
nodes using BERT representations. By jointly training the BERT and GCN modules
within BertGCN, the proposed model is able to leverage the advantages of both
worlds: large-scale pretraining which takes the advantage of the massive amount
of raw data and transductive learning which jointly learns representations for
both training data and unlabeled test data by propagating label influence
through graph convolution. Experiments show that BertGCN achieves SOTA
performances on a wide range of text classification datasets. Code is available
at https://github.com/ZeroRin/BertGCN.
| 2,022 |
Computation and Language
|
Encoding Explanatory Knowledge for Zero-shot Science Question Answering
|
This paper describes N-XKT (Neural encoding based on eXplanatory Knowledge
Transfer), a novel method for the automatic transfer of explanatory knowledge
through neural encoding mechanisms. We demonstrate that N-XKT is able to
improve accuracy and generalization on science Question Answering (QA).
Specifically, by leveraging facts from background explanatory knowledge
corpora, the N-XKT model shows a clear improvement on zero-shot QA.
Furthermore, we show that N-XKT can be fine-tuned on a target QA dataset,
enabling faster convergence and more accurate results. A systematic analysis is
conducted to quantitatively analyze the performance of the N-XKT model and the
impact of different categories of knowledge on the zero-shot generalization
task.
| 2,021 |
Computation and Language
|
Building a Question and Answer System for News Domain
|
This project attempts to build a Question- Answering system in the News
Domain, where Passages will be News articles, and anyone can ask a Question
against it. We have built a span-based model using an Attention mechanism,
where the model predicts the answer to a question as to the position of the
start and end tokens in a paragraph. For training our model, we have used the
Stanford Question and Answer (SQuAD 2.0) dataset[1]. To do well on SQuAD 2.0,
systems must not only answer questions when possible but also determine when no
answer is supported by the paragraph and abstain from answering. Our model
architecture comprises three layers- Embedding Layer, RNN Layer, and the
Attention Layer. For the Embedding layer, we used GloVe and the Universal
Sentence Encoder. For the RNN Layer, we built variations of the RNN Layer
including bi-LSTM and Stacked LSTM and we built an Attention Layer using a
Context to Question Attention and also improvised on the innovative
Bidirectional Attention Layer. Our best performing model which uses GloVe
Embedding combined with Bi-LSTM and Context to Question Attention achieved an
F1 Score and EM of 33.095 and 33.094 respectively. We also leveraged transfer
learning and built a Transformer based model using BERT. The BERT-based model
achieved an F1 Score and EM of 57.513 and 49.769 respectively. We concluded
that the BERT model is superior in all aspects of answering various types of
questions.
| 2,021 |
Computation and Language
|
Conversational Negation using Worldly Context in Compositional
Distributional Semantics
|
We propose a framework to model an operational conversational negation by
applying worldly context (prior knowledge) to logical negation in compositional
distributional semantics. Given a word, our framework can create its negation
that is similar to how humans perceive negation. The framework corrects logical
negation to weight meanings closer in the entailment hierarchy more than
meanings further apart. The proposed framework is flexible to accommodate
different choices of logical negations, compositions, and worldly context
generation. In particular, we propose and motivate a new logical negation using
matrix inverse.
We validate the sensibility of our conversational negation framework by
performing experiments, leveraging density matrices to encode graded entailment
information. We conclude that the combination of subtraction negation and
phaser in the basis of the negated word yields the highest Pearson correlation
of 0.635 with human ratings.
| 2,021 |
Computation and Language
|
Stacked Acoustic-and-Textual Encoding: Integrating the Pre-trained
Models into Speech Translation Encoders
|
Encoder pre-training is promising in end-to-end Speech Translation (ST),
given the fact that speech-to-translation data is scarce. But ST encoders are
not simple instances of Automatic Speech Recognition (ASR) or Machine
Translation (MT) encoders. For example, we find that ASR encoders lack the
global context representation, which is necessary for translation, whereas MT
encoders are not designed to deal with long but locally attentive acoustic
sequences. In this work, we propose a Stacked Acoustic-and-Textual Encoding
(SATE) method for speech translation. Our encoder begins with processing the
acoustic sequence as usual, but later behaves more like an MT encoder for a
global representation of the input sequence. In this way, it is straightforward
to incorporate the pre-trained models into the system. Also, we develop an
adaptor module to alleviate the representation inconsistency between the
pre-trained ASR encoder and MT encoder, and develop a multi-teacher knowledge
distillation method to preserve the pre-training knowledge. Experimental
results on the LibriSpeech En-Fr and MuST-C En-De ST tasks show that our method
achieves state-of-the-art BLEU scores of 18.3 and 25.2. To our knowledge, we
are the first to develop an end-to-end ST system that achieves comparable or
even better BLEU performance than the cascaded ST counterpart when large-scale
ASR and MT data is available.
| 2,021 |
Computation and Language
|
The Semantic Brand Score
|
The Semantic Brand Score (SBS) is a new measure of brand importance
calculated on text data, combining methods of social network and semantic
analysis. This metric is flexible as it can be used in different contexts and
across products, markets and languages. It is applicable not only to brands,
but also to multiple sets of words. The SBS, described together with its three
dimensions of brand prevalence, diversity and connectivity, represents a
contribution to the research on brand equity and on word co-occurrence
networks. It can be used to support decision-making processes within companies;
for example, it can be applied to forecast a company's stock price or to assess
brand importance with respect to competitors. On the one side, the SBS relates
to familiar constructs of brand equity, on the other, it offers new
perspectives for effective strategic management of brands in the era of big
data.
| 2,018 |
Computation and Language
|
The Greedy and Recursive Search for Morphological Productivity
|
As children acquire the knowledge of their language's morphology, they
invariably discover the productive processes that can generalize to new words.
Morphological learning is made challenging by the fact that even fully
productive rules have exceptions, as in the well-known case of English past
tense verbs, which features the -ed rule against the irregular verbs. The
Tolerance Principle is a recent proposal that provides a precise threshold of
exceptions that a productive rule can withstand. Its empirical application so
far, however, requires the researcher to fully specify rules defined over a set
of words. We propose a greedy search model that automatically hypothesizes
rules and evaluates their productivity over a vocabulary. When the search for
broader productivity fails, the model recursively subdivides the vocabulary and
continues the search for productivity over narrower rules. Trained on
psychologically realistic data from child-directed input, our model displays
developmental patterns observed in child morphology acquisition, including the
notoriously complex case of German noun pluralization. It also produces
responses to nonce words that, despite receiving only a fraction of the
training data, are more similar to those of human subjects than current neural
network models' responses are.
| 2,021 |
Computation and Language
|
Kleister: Key Information Extraction Datasets Involving Long Documents
with Complex Layouts
|
The relevance of the Key Information Extraction (KIE) task is increasingly
important in natural language processing problems. But there are still only a
few well-defined problems that serve as benchmarks for solutions in this area.
To bridge this gap, we introduce two new datasets (Kleister NDA and Kleister
Charity). They involve a mix of scanned and born-digital long formal
English-language documents. In these datasets, an NLP system is expected to
find or infer various types of entities by employing both textual and
structural layout features. The Kleister Charity dataset consists of 2,788
annual financial reports of charity organizations, with 61,643 unique pages and
21,612 entities to extract. The Kleister NDA dataset has 540 Non-disclosure
Agreements, with 3,229 unique pages and 2,160 entities to extract. We provide
several state-of-the-art baseline systems from the KIE domain (Flair, BERT,
RoBERTa, LayoutLM, LAMBERT), which show that our datasets pose a strong
challenge to existing models. The best model achieved an 81.77% and an 83.57%
F1-score on respectively the Kleister NDA and the Kleister Charity datasets. We
share the datasets to encourage progress on more in-depth and complex
information extraction tasks.
| 2,021 |
Computation and Language
|
Playing Codenames with Language Graphs and Word Embeddings
|
Although board games and video games have been studied for decades in
artificial intelligence research, challenging word games remain relatively
unexplored. Word games are not as constrained as games like chess or poker.
Instead, word game strategy is defined by the players' understanding of the way
words relate to each other. The word game Codenames provides a unique
opportunity to investigate common sense understanding of relationships between
words, an important open challenge. We propose an algorithm that can generate
Codenames clues from the language graph BabelNet or from any of several
embedding methods - word2vec, GloVe, fastText or BERT. We introduce a new
scoring function that measures the quality of clues, and we propose a weighting
term called DETECT that incorporates dictionary-based word representations and
document frequency to improve clue selection. We develop BabelNet-Word
Selection Framework (BabelNet-WSF) to improve BabelNet clue quality and
overcome the computational barriers that previously prevented leveraging
language graphs for Codenames. Extensive experiments with human evaluators
demonstrate that our proposed innovations yield state-of-the-art performance,
with up to 102.8% improvement in precision@2 in some cases. Overall, this work
advances the formal study of word games and approaches for common sense
language understanding.
| 2,021 |
Computation and Language
|
Black or White but never neutral: How readers perceive identity from
yellow or skin-toned emoji
|
Research in sociology and linguistics shows that people use language not only
to express their own identity but to understand the identity of others. Recent
work established a connection between expression of identity and emoji usage on
social media, through use of emoji skin tone modifiers. Motivated by that
finding, this work asks if, as with language, readers are sensitive to such
acts of self-expression and use them to understand the identity of authors. In
behavioral experiments (n=488), where text and emoji content of social media
posts were carefully controlled before being presented to participants, we find
in the affirmative -- emoji are a salient signal of author identity. That
signal is distinct from, and complementary to, the one encoded in language.
Participant groups (based on self-identified ethnicity) showed no differences
in how they perceive this signal, except in the case of the default yellow
emoji. While both groups associate this with a White identity, the effect was
stronger in White participants. Our finding that emoji can index social
variables will have experimental applications for researchers but also
implications for designers: supposedly ``neutral`` defaults may be more
representative of some users than others.
| 2,021 |
Computation and Language
|
MATE-KD: Masked Adversarial TExt, a Companion to Knowledge Distillation
|
The advent of large pre-trained language models has given rise to rapid
progress in the field of Natural Language Processing (NLP). While the
performance of these models on standard benchmarks has scaled with size,
compression techniques such as knowledge distillation have been key in making
them practical. We present, MATE-KD, a novel text-based adversarial training
algorithm which improves the performance of knowledge distillation. MATE-KD
first trains a masked language model based generator to perturb text by
maximizing the divergence between teacher and student logits. Then using
knowledge distillation a student is trained on both the original and the
perturbed training samples. We evaluate our algorithm, using BERT-based models,
on the GLUE benchmark and demonstrate that MATE-KD outperforms competitive
adversarial learning and data augmentation baselines. On the GLUE test set our
6 layer RoBERTa based model outperforms BERT-Large.
| 2,021 |
Computation and Language
|
Go Beyond Plain Fine-tuning: Improving Pretrained Models for Social
Commonsense
|
Pretrained language models have demonstrated outstanding performance in many
NLP tasks recently. However, their social intelligence, which requires
commonsense reasoning about the current situation and mental states of others,
is still developing. Towards improving language models' social intelligence, we
focus on the Social IQA dataset, a task requiring social and emotional
commonsense reasoning. Building on top of the pretrained RoBERTa and GPT2
models, we propose several architecture variations and extensions, as well as
leveraging external commonsense corpora, to optimize the model for Social IQA.
Our proposed system achieves competitive results as those top-ranking models on
the leaderboard. This work demonstrates the strengths of pretrained language
models, and provides viable ways to improve their performance for a particular
task.
| 2,021 |
Computation and Language
|
Better than BERT but Worse than Baseline
|
This paper compares BERT-SQuAD and Ab3P on the Abbreviation Definition
Identification (ADI) task. ADI inputs a text and outputs short forms
(abbreviations/acronyms) and long forms (expansions). BERT with reranking
improves over BERT without reranking but fails to reach the Ab3P rule-based
baseline. What is BERT missing? Reranking introduces two new features:
charmatch and freq. The first feature identifies opportunities to take
advantage of character constraints in acronyms and the second feature
identifies opportunities to take advantage of frequency constraints across
documents.
| 2,021 |
Computation and Language
|
Analysing The Impact Of Linguistic Features On Cross-Lingual Transfer
|
There is an increasing amount of evidence that in cases with little or no
data in a target language, training on a different language can yield
surprisingly good results. However, currently there are no established
guidelines for choosing the training (source) language. In attempt to solve
this issue we thoroughly analyze a state-of-the-art multilingual model and try
to determine what impacts good transfer between languages. As opposed to the
majority of multilingual NLP literature, we don't only train on English, but on
a group of almost 30 languages. We show that looking at particular syntactic
features is 2-4 times more helpful in predicting the performance than an
aggregated syntactic similarity. We find out that the importance of syntactic
features strongly differs depending on the downstream task - no single feature
is a good performance predictor for all NLP tasks. As a result, one should not
expect that for a target language $L_1$ there is a single language $L_2$ that
is the best choice for any NLP task (for instance, for Bulgarian, the best
source language is French on POS tagging, Russian on NER and Thai on NLI). We
discuss the most important linguistic features affecting the transfer quality
using statistical and machine learning methods.
| 2,021 |
Computation and Language
|
Spelling Correction with Denoising Transformer
|
We present a novel method of performing spelling correction on short input
strings, such as search queries or individual words. At its core lies a
procedure for generating artificial typos which closely follow the error
patterns manifested by humans. This procedure is used to train the production
spelling correction model based on a transformer architecture. This model is
currently served in the HubSpot product search. We show that our approach to
typo generation is superior to the widespread practice of adding noise, which
ignores human patterns. We also demonstrate how our approach may be extended to
resource-scarce settings and train spelling correction models for Arabic,
Greek, Russian, and Setswana languages, without using any labeled data.
| 2,021 |
Computation and Language
|
Multilingual Offensive Language Identification for Low-resource
Languages
|
Offensive content is pervasive in social media and a reason for concern to
companies and government organizations. Several studies have been recently
published investigating methods to detect the various forms of such content
(e.g. hate speech, cyberbullying, and cyberaggression). The clear majority of
these studies deal with English partially because most annotated datasets
available contain English data. In this paper, we take advantage of available
English datasets by applying cross-lingual contextual word embeddings and
transfer learning to make predictions in low-resource languages. We project
predictions on comparable data in Arabic, Bengali, Danish, Greek, Hindi,
Spanish, and Turkish. We report results of 0.8415 F1 macro for Bengali in
TRAC-2 shared task, 0.8532 F1 macro for Danish and 0.8701 F1 macro for Greek in
OffensEval 2020, 0.8568 F1 macro for Hindi in HASOC 2019 shared task and 0.7513
F1 macro for Spanish in in SemEval-2019 Task 5 (HatEval) showing that our
approach compares favourably to the best systems submitted to recent shared
tasks on these three languages. Additionally, we report competitive performance
on Arabic, and Turkish using the training and development sets of OffensEval
2020 shared task. The results for all languages confirm the robustness of
cross-lingual contextual embeddings and transfer learning for this task.
| 2,021 |
Computation and Language
|
Designing Multimodal Datasets for NLP Challenges
|
In this paper, we argue that the design and development of multimodal
datasets for natural language processing (NLP) challenges should be enhanced in
two significant respects: to more broadly represent commonsense semantic
inferences; and to better reflect the dynamics of actions and events, through a
substantive alignment of textual and visual information. We identify challenges
and tasks that are reflective of linguistic and cognitive competencies that
humans have when speaking and reasoning, rather than merely the performance of
systems on isolated tasks. We introduce the distinction between challenge-based
tasks and competence-based performance, and describe a diagnostic dataset,
Recipe-to-Video Questions (R2VQ), designed for testing competence-based
comprehension over a multimodal recipe collection (http://r2vq.org/). The
corpus contains detailed annotation supporting such inferencing tasks and
facilitating a rich set of question families that we use to evaluate NLP
systems.
| 2,021 |
Computation and Language
|
Are Larger Pretrained Language Models Uniformly Better? Comparing
Performance at the Instance Level
|
Larger language models have higher accuracy on average, but are they better
on every single instance (datapoint)? Some work suggests larger models have
higher out-of-distribution robustness, while other work suggests they have
lower accuracy on rare subgroups. To understand these differences, we
investigate these models at the level of individual instances. However, one
major challenge is that individual predictions are highly sensitive to noise in
the randomness in training. We develop statistically rigorous methods to
address this, and after accounting for pretraining and finetuning noise, we
find that our BERT-Large is worse than BERT-Mini on at least 1-4% of instances
across MNLI, SST-2, and QQP, compared to the overall accuracy improvement of
2-10%. We also find that finetuning noise increases with model size and that
instance-level accuracy has momentum: improvement from BERT-Mini to BERT-Medium
correlates with improvement from BERT-Medium to BERT-Large. Our findings
suggest that instance-level predictions provide a rich source of information;
we therefore, recommend that researchers supplement model weights with model
predictions.
| 2,021 |
Computation and Language
|
Towards Human-Free Automatic Quality Evaluation of German Summarization
|
Evaluating large summarization corpora using humans has proven to be
expensive from both the organizational and the financial perspective.
Therefore, many automatic evaluation metrics have been developed to measure the
summarization quality in a fast and reproducible way. However, most of the
metrics still rely on humans and need gold standard summaries generated by
linguistic experts. Since BLANC does not require golden summaries and
supposedly can use any underlying language model, we consider its application
to the evaluation of summarization in German. This work demonstrates how to
adjust the BLANC metric to a language other than English. We compare BLANC
scores with the crowd and expert ratings, as well as with commonly used
automatic metrics on a German summarization data set. Our results show that
BLANC in German is especially good in evaluating informativeness.
| 2,021 |
Computation and Language
|
HyKnow: End-to-End Task-Oriented Dialog Modeling with Hybrid Knowledge
Management
|
Task-oriented dialog (TOD) systems typically manage structured knowledge
(e.g. ontologies and databases) to guide the goal-oriented conversations.
However, they fall short of handling dialog turns grounded on unstructured
knowledge (e.g. reviews and documents). In this paper, we formulate a task of
modeling TOD grounded on both structured and unstructured knowledge. To address
this task, we propose a TOD system with hybrid knowledge management, HyKnow. It
extends the belief state to manage both structured and unstructured knowledge,
and is the first end-to-end model that jointly optimizes dialog modeling
grounded on these two kinds of knowledge. We conduct experiments on the
modified version of MultiWOZ 2.1 dataset, where dialogs are grounded on hybrid
knowledge. Experimental results show that HyKnow has strong end-to-end
performance compared to existing TOD systems. It also outperforms the pipeline
knowledge management schemes, with higher unstructured knowledge retrieval
accuracy.
| 2,021 |
Computation and Language
|
Semi-Supervised Variational Reasoning for Medical Dialogue Generation
|
Medical dialogue generation aims to provide automatic and accurate responses
to assist physicians to obtain diagnosis and treatment suggestions in an
efficient manner. In medical dialogues two key characteristics are relevant for
response generation: patient states (such as symptoms, medication) and
physician actions (such as diagnosis, treatments). In medical scenarios
large-scale human annotations are usually not available, due to the high costs
and privacy requirements. Hence, current approaches to medical dialogue
generation typically do not explicitly account for patient states and physician
actions, and focus on implicit representation instead. We propose an end-to-end
variational reasoning approach to medical dialogue generation. To be able to
deal with a limited amount of labeled data, we introduce both patient state and
physician action as latent variables with categorical priors for explicit
patient state tracking and physician policy learning, respectively. We propose
a variational Bayesian generative approach to approximate posterior
distributions over patient states and physician actions. We use an efficient
stochastic gradient variational Bayes estimator to optimize the derived
evidence lower bound, where a 2-stage collapsed inference method is proposed to
reduce the bias during model training. A physician policy network composed of
an action-classifier and two reasoning detectors is proposed for augmented
reasoning ability. We conduct experiments on three datasets collected from
medical platforms. Our experimental results show that the proposed method
outperforms state-of-the-art baselines in terms of objective and subjective
evaluation metrics. Our experiments also indicate that our proposed
semi-supervised reasoning method achieves a comparable performance as
state-of-the-art fully supervised learning baselines for physician policy
learning.
| 2,021 |
Computation and Language
|
Thematic Fit Bits: Annotation Quality and Quantity Interplay for Event
Participant Representation
|
Modeling thematic fit (a verb--argument compositional semantics task)
currently requires a very large burden of labeled data. We take a
linguistically machine-annotated large corpus and replace corpus layers with
output from higher-quality, more modern taggers. We compare the old and new
corpus versions' impact on a verb--argument fit modeling task, using a
high-performing neural approach. We discover that higher annotation quality
dramatically reduces our data requirement while demonstrating better supervised
predicate-argument classification. But in applying the model to
psycholinguistic tasks outside the training objective, we see clear gains at
scale, but only in one of two thematic fit estimation tasks, and no clear gains
on the other. We also see that quality improves with training size, but perhaps
plateauing or even declining in one task. Last, we tested the effect of role
set size. All this suggests that the quality/quantity interplay is not all you
need. We replicate previous studies while modifying certain role representation
details and set a new state-of-the-art in event modeling, using a fraction of
the data. We make the new corpus version public.
| 2,022 |
Computation and Language
|
Linguistic Inspired Graph Analysis
|
Isomorphisms allow human cognition to transcribe a potentially unsolvable
problem from one domain to a different domain where the problem might be more
easily addressed. Current approaches only focus on transcribing structural
information from the source to target structure, ignoring semantic and
pragmatic information. Functional Language Theory presents five subconstructs
for the classification and understanding of languages. By deriving a mapping
between the metamodels in linguistics and graph theory it will be shown that
currently, no constructs exist in canonical graphs for the representation of
semantic and pragmatic information. It is found that further work needs to be
done to understand how graphs can be enriched to allow for isomorphisms to
capture semantic and pragmatic information. This capturing of additional
information could lead to understandings of the source structure and enhanced
manipulations and interrogations of the contained relationships. Current
mathematical graph structures in their general definition do not allow for the
expression of higher information levels of a source.
| 2,021 |
Computation and Language
|
Retrieval-Free Knowledge-Grounded Dialogue Response Generation with
Adapters
|
To diversify and enrich generated dialogue responses, knowledge-grounded
dialogue has been investigated in recent years. The existing methods tackle the
knowledge grounding challenge by retrieving the relevant sentences over a large
corpus and augmenting the dialogues with explicit extra information. Despite
their success, however, the existing works have drawbacks in inference
efficiency. This paper proposes KnowExpert, a framework to bypass the explicit
retrieval process and inject knowledge into the pre-trained language models
with lightweight adapters and adapt to the knowledge-grounded dialogue task. To
the best of our knowledge, this is the first attempt to tackle this challenge
without retrieval in this task under an open-domain chit-chat scenario. The
experimental results show that Knowexpert performs comparably with some
retrieval-based baselines while being time-efficient in inference,
demonstrating the effectiveness of our proposed method.
| 2,022 |
Computation and Language
|
Video Corpus Moment Retrieval with Contrastive Learning
|
Given a collection of untrimmed and unsegmented videos, video corpus moment
retrieval (VCMR) is to retrieve a temporal moment (i.e., a fraction of a video)
that semantically corresponds to a given text query. As video and text are from
two distinct feature spaces, there are two general approaches to address VCMR:
(i) to separately encode each modality representations, then align the two
modality representations for query processing, and (ii) to adopt fine-grained
cross-modal interaction to learn multi-modal representations for query
processing. While the second approach often leads to better retrieval accuracy,
the first approach is far more efficient. In this paper, we propose a Retrieval
and Localization Network with Contrastive Learning (ReLoCLNet) for VCMR. We
adopt the first approach and introduce two contrastive learning objectives to
refine video encoder and text encoder to learn video and text representations
separately but with better alignment for VCMR. The video contrastive learning
(VideoCL) is to maximize mutual information between query and candidate video
at video-level. The frame contrastive learning (FrameCL) aims to highlight the
moment region corresponds to the query at frame-level, within a video.
Experimental results show that, although ReLoCLNet encodes text and video
separately for efficiency, its retrieval accuracy is comparable with baselines
adopting cross-modal interaction learning.
| 2,021 |
Computation and Language
|
Predicting Text Readability from Scrolling Interactions
|
Judging the readability of text has many important applications, for instance
when performing text simplification or when sourcing reading material for
language learners. In this paper, we present a 518 participant study which
investigates how scrolling behaviour relates to the readability of a text. We
make our dataset publicly available and show that (1) there are statistically
significant differences in the way readers interact with text depending on the
text level, (2) such measures can be used to predict the readability of text,
and (3) the background of a reader impacts their reading interactions and the
factors contributing to text difficulty.
| 2,021 |
Computation and Language
|
SaRoCo: Detecting Satire in a Novel Romanian Corpus of News Articles
|
In this work, we introduce a corpus for satire detection in Romanian news. We
gathered 55,608 public news articles from multiple real and satirical news
sources, composing one of the largest corpora for satire detection regardless
of language and the only one for the Romanian language. We provide an official
split of the text samples, such that training news articles belong to different
sources than test news articles, thus ensuring that models do not achieve high
performance simply due to overfitting. We conduct experiments with two
state-of-the-art deep neural models, resulting in a set of strong baselines for
our novel corpus. Our results show that the machine-level accuracy for satire
detection in Romanian is quite low (under 73% on the test set) compared to the
human-level accuracy (87%), leaving enough room for improvement in future
research.
| 2,021 |
Computation and Language
|
Conversational AI Systems for Social Good: Opportunities and Challenges
|
Conversational artificial intelligence (ConvAI) systems have attracted much
academic and commercial attention recently, making significant progress on both
fronts. However, little existing work discusses how these systems can be
developed and deployed for social good in real-world applications, with
comprehensive case studies and analyses of pros and cons. In this paper, we
briefly review the progress the community has made towards better ConvAI
systems and reflect on how existing technologies can help advance social good
initiatives from various angles that are unique for ConvAI, or not yet become
common knowledge in the community. We further discuss about the challenges
ahead for ConvAI systems to better help us achieve these goals and highlight
the risks involved in their development and deployment in the real world.
| 2,022 |
Computation and Language
|
NLP is Not enough -- Contextualization of User Input in Chatbots
|
AI chatbots have made vast strides in technology improvement in recent years
and are already operational in many industries. Advanced Natural Language
Processing techniques, based on deep networks, efficiently process user
requests to carry out their functions. As chatbots gain traction, their
applicability in healthcare is an attractive proposition due to the reduced
economic and people costs of an overburdened system. However, healthcare bots
require safe and medically accurate information capture, which deep networks
aren't yet capable of due to user text and speech variations. Knowledge in
symbolic structures is more suited for accurate reasoning but cannot handle
natural language processing directly. Thus, in this paper, we study the effects
of combining knowledge and neural representations on chatbot safety, accuracy,
and understanding.
| 2,021 |
Computation and Language
|
Distilling BERT for low complexity network training
|
This paper studies the efficiency of transferring BERT learnings to low
complexity models like BiLSTM, BiLSTM with attention and shallow CNNs using
sentiment analysis on SST-2 dataset. It also compares the complexity of
inference of the BERT model with these lower complexity models and underlines
the importance of these techniques in enabling high performance NLP models on
edge devices like mobiles, tablets and MCU development boards like Raspberry Pi
etc. and enabling exciting new applications.
| 2,021 |
Computation and Language
|
Shades of confusion: Lexical uncertainty modulates ad hoc coordination
in an interactive communication task
|
There is substantial variability in the expectations that communication
partners bring into interactions, creating the potential for misunderstandings.
To directly probe these gaps and our ability to overcome them, we propose a
communication task based on color-concept associations. In Experiment 1, we
establish several key properties of the mental representations of these
expectations, or lexical priors, based on recent probabilistic theories.
Associations are more variable for abstract concepts, variability is
represented as uncertainty within each individual, and uncertainty enables
accurate predictions about whether others are likely to share the same
association. In Experiment 2, we then examine the downstream consequences of
these representations for communication. Accuracy is initially low when
communicating about concepts with more variable associations, but rapidly
increases as participants form ad hoc conventions. Together, our findings
suggest that people cope with variability by maintaining well-calibrated
uncertainty about their partner and appropriately adaptable representations of
their own.
| 2,022 |
Computation and Language
|
RetGen: A Joint framework for Retrieval and Grounded Text Generation
Modeling
|
Recent advances in large-scale pre-training such as GPT-3 allow seemingly
high quality text to be generated from a given prompt. However, such generation
systems often suffer from problems of hallucinated facts, and are not
inherently designed to incorporate useful external information. Grounded
generation models appear to offer remedies, but their training typically relies
on rarely-available parallel data where information-relevant documents are
provided for context. We propose a framework that alleviates this data
constraint by jointly training a grounded generator and document retriever on
the language model signal. The model learns to reward retrieval of the
documents with the highest utility in generation, and attentively combines them
using a Mixture-of-Experts (MoE) ensemble to generate follow-on text. We
demonstrate that both generator and retriever can take advantage of this joint
training and work synergistically to produce more informative and relevant text
in both prose and dialogue generation.
| 2,022 |
Computation and Language
|
Adversarial Learning for Zero-Shot Stance Detection on Social Media
|
Stance detection on social media can help to identify and understand slanted
news or commentary in everyday life. In this work, we propose a new model for
zero-shot stance detection on Twitter that uses adversarial learning to
generalize across topics. Our model achieves state-of-the-art performance on a
number of unseen test topics with minimal computational costs. In addition, we
extend zero-shot stance detection to new topics, highlighting future directions
for zero-shot transfer.
| 2,021 |
Computation and Language
|
Dynamic Multi-Branch Layers for On-Device Neural Machine Translation
|
With the rapid development of artificial intelligence (AI), there is a trend
in moving AI applications, such as neural machine translation (NMT), from cloud
to mobile devices. Constrained by limited hardware resources and battery, the
performance of on-device NMT systems is far from satisfactory. Inspired by
conditional computation, we propose to improve the performance of on-device NMT
systems with dynamic multi-branch layers. Specifically, we design a layer-wise
dynamic multi-branch network with only one branch activated during training and
inference. As not all branches are activated during training, we propose
shared-private reparameterization to ensure sufficient training for each
branch. At almost the same computational cost, our method achieves improvements
of up to 1.7 BLEU points on the WMT14 English-German translation task and 1.8
BLEU points on the WMT20 Chinese-English translation task over the Transformer
model, respectively. Compared with a strong baseline that also uses multiple
branches, the proposed method is up to 1.5 times faster with the same number of
parameters.
| 2,022 |
Computation and Language
|
DaLAJ - a dataset for linguistic acceptability judgments for Swedish:
Format, baseline, sharing
|
We present DaLAJ 1.0, a Dataset for Linguistic Acceptability Judgments for
Swedish, comprising 9 596 sentences in its first version; and the initial
experiment using it for the binary classification task. DaLAJ is based on the
SweLL second language learner data, consisting of essays at different levels of
proficiency. To make sure the dataset can be freely available despite the GDPR
regulations, we have sentence-scrambled learner essays and removed part of the
metadata about learners, keeping for each sentence only information about the
mother tongue and the level of the course where the essay has been written. We
use the normalized version of learner language as the basis for the DaLAJ
sentences, and keep only one error per sentence. We repeat the same sentence
for each individual correction tag used in the sentence. For DaLAJ 1.0 we have
used four error categories (out of 35 available in SweLL), all connected to
lexical or word-building choices. Our baseline results for the binary
classification show an accuracy of 58% for DaLAJ 1.0 using BERT embeddings. The
dataset is included in the SwedishGlue (Swe. SuperLim) benchmark. Below, we
describe the format of the dataset, first experiments, our insights and the
motivation for the chosen approach to data sharing.
| 2,021 |
Computation and Language
|
Out-of-Manifold Regularization in Contextual Embedding Space for Text
Classification
|
Recent studies on neural networks with pre-trained weights (i.e., BERT) have
mainly focused on a low-dimensional subspace, where the embedding vectors
computed from input words (or their contexts) are located. In this work, we
propose a new approach to finding and regularizing the remainder of the space,
referred to as out-of-manifold, which cannot be accessed through the words.
Specifically, we synthesize the out-of-manifold embeddings based on two
embeddings obtained from actually-observed words, to utilize them for
fine-tuning the network. A discriminator is trained to detect whether an input
embedding is located inside the manifold or not, and simultaneously, a
generator is optimized to produce new embeddings that can be easily identified
as out-of-manifold by the discriminator. These two modules successfully
collaborate in a unified and end-to-end manner for regularizing the
out-of-manifold. Our extensive evaluation on various text classification
benchmarks demonstrates the effectiveness of our approach, as well as its good
compatibility with existing data augmentation techniques which aim to enhance
the manifold.
| 2,021 |
Computation and Language
|
Classifying Long Clinical Documents with Pre-trained Transformers
|
Automatic phenotyping is a task of identifying cohorts of patients that match
a predefined set of criteria. Phenotyping typically involves classifying long
clinical documents that contain thousands of tokens. At the same time, recent
state-of-art transformer-based pre-trained language models limit the input to a
few hundred tokens (e.g. 512 tokens for BERT). We evaluate several strategies
for incorporating pre-trained sentence encoders into document-level
representations of clinical text, and find that hierarchical transformers
without pre-training are competitive with task pre-trained models.
| 2,021 |
Computation and Language
|
DialogSum: A Real-Life Scenario Dialogue Summarization Dataset
|
Proposal of large-scale datasets has facilitated research on deep neural
models for news summarization. Deep learning can also be potentially useful for
spoken dialogue summarization, which can benefit a range of real-life scenarios
including customer service management and medication tracking. To this end, we
propose DialogSum, a large-scale labeled dialogue summarization dataset. We
conduct empirical analysis on DialogSum using state-of-the-art neural
summarizers. Experimental results show unique challenges in dialogue
summarization, such as spoken terms, special discourse structures, coreferences
and ellipsis, pragmatics and social common sense, which require specific
representation learning technologies to better deal with.
| 2,021 |
Computation and Language
|
Locate and Label: A Two-stage Identifier for Nested Named Entity
Recognition
|
Named entity recognition (NER) is a well-studied task in natural language
processing. Traditional NER research only deals with flat entities and ignores
nested entities. The span-based methods treat entity recognition as a span
classification task. Although these methods have the innate ability to handle
nested NER, they suffer from high computational cost, ignorance of boundary
information, under-utilization of the spans that partially match with entities,
and difficulties in long entity recognition. To tackle these issues, we propose
a two-stage entity identifier. First we generate span proposals by filtering
and boundary regression on the seed spans to locate the entities, and then
label the boundary-adjusted span proposals with the corresponding categories.
Our method effectively utilizes the boundary information of entities and
partially matched spans during training. Through boundary regression, entities
of any length can be covered theoretically, which improves the ability to
recognize long entities. In addition, many low-quality seed spans are filtered
out in the first stage, which reduces the time complexity of inference.
Experiments on nested NER datasets demonstrate that our proposed method
outperforms previous state-of-the-art models.
| 2,021 |
Computation and Language
|
A cost-benefit analysis of cross-lingual transfer methods
|
An effective method for cross-lingual transfer is to fine-tune a bilingual or
multilingual model on a supervised dataset in one language and evaluating it on
another language in a zero-shot manner. Translating examples at training time
or inference time are also viable alternatives. However, there are costs
associated with these methods that are rarely addressed in the literature. In
this work, we analyze cross-lingual methods in terms of their effectiveness
(e.g., accuracy), development and deployment costs, as well as their latencies
at inference time. Our experiments on three tasks indicate that the best
cross-lingual method is highly task-dependent. Finally, by combining zero-shot
and translation methods, we achieve the state-of-the-art in two of the three
datasets used in this work. Based on these results, we question the need for
manually labeled training data in a target language. Code and translated
datasets are available at https://github.com/unicamp-dl/cross-lingual-analysis
| 2,021 |
Computation and Language
|
Empathetic Dialog Generation with Fine-Grained Intents
|
Empathetic dialog generation aims at generating coherent responses following
previous dialog turns and, more importantly, showing a sense of caring and a
desire to help. Existing models either rely on pre-defined emotion labels to
guide the response generation, or use deterministic rules to decide the emotion
of the response. With the advent of advanced language models, it is possible to
learn subtle interactions directly from the dataset, providing that the emotion
categories offer sufficient nuances and other non-emotional but emotional
regulating intents are included. In this paper, we describe how to incorporate
a taxonomy of 32 emotion categories and 8 additional emotion regulating intents
to succeed the task of empathetic response generation. To facilitate the
training, we also curated a large-scale emotional dialog dataset from movie
subtitles. Through a carefully designed crowdsourcing experiment, we evaluated
and demonstrated how our model produces more empathetic dialogs compared with
its baselines.
| 2,021 |
Computation and Language
|
Towards Navigation by Reasoning over Spatial Configurations
|
We deal with the navigation problem where the agent follows natural language
instructions while observing the environment. Focusing on language
understanding, we show the importance of spatial semantics in grounding
navigation instructions into visual perceptions. We propose a neural agent that
uses the elements of spatial configurations and investigate their influence on
the navigation agent's reasoning ability. Moreover, we model the sequential
execution order and align visual objects with spatial configurations in the
instruction. Our neural agent improves strong baselines on the seen
environments and shows competitive performance on the unseen environments.
Additionally, the experimental results demonstrate that explicit modeling of
spatial semantic elements in the instructions can improve the grounding and
spatial reasoning of the model.
| 2,021 |
Computation and Language
|
QAConv: Question Answering on Informative Conversations
|
This paper introduces QAConv, a new question answering (QA) dataset that uses
conversations as a knowledge source. We focus on informative conversations,
including business emails, panel discussions, and work channels. Unlike
open-domain and task-oriented dialogues, these conversations are usually long,
complex, asynchronous, and involve strong domain knowledge. In total, we
collect 34,608 QA pairs from 10,259 selected conversations with both
human-written and machine-generated questions. We use a question generator and
a dialogue summarizer as auxiliary tools to collect and recommend questions.
The dataset has two testing scenarios: chunk mode and full mode, depending on
whether the grounded partial conversation is provided or retrieved.
Experimental results show that state-of-the-art pretrained QA systems have
limited zero-shot performance and tend to predict our questions as
unanswerable. Our dataset provides a new training and evaluation testbed to
facilitate QA on conversations research.
| 2,022 |
Computation and Language
|
Thank you BART! Rewarding Pre-Trained Models Improves Formality Style
Transfer
|
Scarcity of parallel data causes formality style transfer models to have
scarce success in preserving content. We show that fine-tuning pre-trained
language (GPT-2) and sequence-to-sequence (BART) models boosts content
preservation, and that this is possible even with limited amounts of parallel
data. Augmenting these models with rewards that target style and content -- the
two core aspects of the task -- we achieve a new state-of-the-art.
| 2,021 |
Computation and Language
|
Plot and Rework: Modeling Storylines for Visual Storytelling
|
Writing a coherent and engaging story is not easy. Creative writers use their
knowledge and worldview to put disjointed elements together to form a coherent
storyline, and work and rework iteratively toward perfection. Automated visual
storytelling (VIST) models, however, make poor use of external knowledge and
iterative generation when attempting to create stories. This paper introduces
PR-VIST, a framework that represents the input image sequence as a story graph
in which it finds the best path to form a storyline. PR-VIST then takes this
path and learns to generate the final story via an iterative training process.
This framework produces stories that are superior in terms of diversity,
coherence, and humanness, per both automatic and human evaluations. An ablation
study shows that both plotting and reworking contribute to the model's
superiority.
| 2,021 |
Computation and Language
|
Counterfactual Interventions Reveal the Causal Effect of Relative Clause
Representations on Agreement Prediction
|
When language models process syntactically complex sentences, do they use
their representations of syntax in a manner that is consistent with the grammar
of the language? We propose AlterRep, an intervention-based method to address
this question. For any linguistic feature of a given sentence, AlterRep
generates counterfactual representations by altering how the feature is
encoded, while leaving intact all other aspects of the original representation.
By measuring the change in a model's word prediction behavior when these
counterfactual representations are substituted for the original ones, we can
draw conclusions about the causal effect of the linguistic feature in question
on the model's behavior. We apply this method to study how BERT models of
different sizes process relative clauses (RCs). We find that BERT variants use
RC boundary information during word prediction in a manner that is consistent
with the rules of English grammar; this RC boundary information generalizes to
a considerable extent across different RC types, suggesting that BERT
represents RCs as an abstract linguistic category.
| 2,021 |
Computation and Language
|
Do Context-Aware Translation Models Pay the Right Attention?
|
Context-aware machine translation models are designed to leverage contextual
information, but often fail to do so. As a result, they inaccurately
disambiguate pronouns and polysemous words that require context for resolution.
In this paper, we ask several questions: What contexts do human translators use
to resolve ambiguous words? Are models paying large amounts of attention to the
same context? What if we explicitly train them to do so? To answer these
questions, we introduce SCAT (Supporting Context for Ambiguous Translations), a
new English-French dataset comprising supporting context words for 14K
translations that professional translators found useful for pronoun
disambiguation. Using SCAT, we perform an in-depth analysis of the context used
to disambiguate, examining positional and lexical characteristics of the
supporting words. Furthermore, we measure the degree of alignment between the
model's attention scores and the supporting context from SCAT, and apply a
guided attention strategy to encourage agreement between the two.
| 2,021 |
Computation and Language
|
EASE: Extractive-Abstractive Summarization with Explanations
|
Current abstractive summarization systems outperform their extractive
counterparts, but their widespread adoption is inhibited by the inherent lack
of interpretability. To achieve the best of both worlds, we propose EASE, an
extractive-abstractive framework for evidence-based text generation and apply
it to document summarization. We present an explainable summarization system
based on the Information Bottleneck principle that is jointly trained for
extraction and abstraction in an end-to-end fashion. Inspired by previous
research that humans use a two-stage framework to summarize long documents
(Jing and McKeown, 2000), our framework first extracts a pre-defined amount of
evidence spans as explanations and then generates a summary using only the
evidence. Using automatic and human evaluations, we show that explanations from
our framework are more relevant than simple baselines, without substantially
sacrificing the quality of the generated summary.
| 2,021 |
Computation and Language
|
BERT Busters: Outlier Dimensions that Disrupt Transformers
|
Multiple studies have shown that Transformers are remarkably robust to
pruning. Contrary to this received wisdom, we demonstrate that pre-trained
Transformer encoders are surprisingly fragile to the removal of a very small
number of features in the layer outputs (<0.0001% of model weights). In case of
BERT and other pre-trained encoder Transformers, the affected component is the
scaling factors and biases in the LayerNorm. The outliers are high-magnitude
normalization parameters that emerge early in pre-training and show up
consistently in the same dimensional position throughout the model. We show
that disabling them significantly degrades both the MLM loss and the downstream
task performance. This effect is observed across several BERT-family models and
other popular pre-trained Transformer architectures, including BART, XLNet and
ELECTRA; we also show a similar effect in GPT-2.
| 2,021 |
Computation and Language
|
The Low-Dimensional Linear Geometry of Contextualized Word
Representations
|
Black-box probing models can reliably extract linguistic features like tense,
number, and syntactic role from pretrained word representations. However, the
manner in which these features are encoded in representations remains poorly
understood. We present a systematic study of the linear geometry of
contextualized word representations in ELMO and BERT. We show that a variety of
linguistic features (including structured dependency relationships) are encoded
in low-dimensional subspaces. We then refine this geometric picture, showing
that there are hierarchical relations between the subspaces encoding general
linguistic categories and more specific ones, and that low-dimensional feature
encodings are distributed rather than aligned to individual neurons. Finally,
we demonstrate that these linear subspaces are causally related to model
behavior, and can be used to perform fine-grained manipulation of BERT's output
distribution.
| 2,021 |
Computation and Language
|
Premise-based Multimodal Reasoning: Conditional Inference on Joint
Textual and Visual Clues
|
It is a common practice for recent works in vision language cross-modal
reasoning to adopt a binary or multi-choice classification formulation taking
as input a set of source image(s) and textual query. In this work, we take a
sober look at such an unconditional formulation in the sense that no prior
knowledge is specified with respect to the source image(s). Inspired by the
designs of both visual commonsense reasoning and natural language inference
tasks, we propose a new task termed Premise-based Multi-modal Reasoning(PMR)
where a textual premise is the background presumption on each source image. The
PMR dataset contains 15,360 manually annotated samples which are created by a
multi-phase crowd-sourcing process. With selected high-quality movie
screenshots and human-curated premise templates from 6 pre-defined categories,
we ask crowd-source workers to write one true hypothesis and three distractors
(4 choices) given the premise and image through a cross-check procedure.
Besides, we generate adversarial samples to alleviate the annotation artifacts
and double the size of PMR. We benchmark various state-of-the-art (pretrained)
multi-modal inference models on PMR and conduct comprehensive experimental
analyses to showcase the utility of our dataset.
| 2,022 |
Computation and Language
|
A Cognitive Regularizer for Language Modeling
|
The uniform information density (UID) hypothesis, which posits that speakers
behaving optimally tend to distribute information uniformly across a linguistic
signal, has gained traction in psycholinguistics as an explanation for certain
syntactic, morphological, and prosodic choices. In this work, we explore
whether the UID hypothesis can be operationalized as an inductive bias for
statistical language modeling. Specifically, we augment the canonical MLE
objective for training language models with a regularizer that encodes UID. In
experiments on ten languages spanning five language families, we find that
using UID regularization consistently improves perplexity in language models,
having a larger effect when training data is limited. Moreover, via an analysis
of generated sequences, we find that UID-regularized language models have other
desirable properties, e.g., they generate text that is more lexically diverse.
Our results not only suggest that UID is a reasonable inductive bias for
language modeling, but also provide an alternative validation of the UID
hypothesis using modern-day NLP tools.
| 2,021 |
Computation and Language
|
Lexicon Enhanced Chinese Sequence Labeling Using BERT Adapter
|
Lexicon information and pre-trained models, such as BERT, have been combined
to explore Chinese sequence labelling tasks due to their respective strengths.
However, existing methods solely fuse lexicon features via a shallow and random
initialized sequence layer and do not integrate them into the bottom layers of
BERT. In this paper, we propose Lexicon Enhanced BERT (LEBERT) for Chinese
sequence labelling, which integrates external lexicon knowledge into BERT
layers directly by a Lexicon Adapter layer. Compared with the existing methods,
our model facilitates deep lexicon knowledge fusion at the lower layers of
BERT. Experiments on ten Chinese datasets of three tasks including Named Entity
Recognition, Word Segmentation, and Part-of-Speech tagging, show that LEBERT
achieves the state-of-the-art results.
| 2,021 |
Computation and Language
|
DirectQE: Direct Pretraining for Machine Translation Quality Estimation
|
Machine Translation Quality Estimation (QE) is a task of predicting the
quality of machine translations without relying on any reference. Recently, the
predictor-estimator framework trains the predictor as a feature extractor,
which leverages the extra parallel corpora without QE labels, achieving
promising QE performance. However, we argue that there are gaps between the
predictor and the estimator in both data quality and training objectives, which
preclude QE models from benefiting from a large number of parallel corpora more
directly. We propose a novel framework called DirectQE that provides a direct
pretraining for QE tasks. In DirectQE, a generator is trained to produce pseudo
data that is closer to the real QE data, and a detector is pretrained on these
data with novel objectives that are akin to the QE task. Experiments on widely
used benchmarks show that DirectQE outperforms existing methods, without using
any pretraining models such as BERT. We also give extensive analyses showing
how fixing the two gaps contributes to our improvements.
| 2,021 |
Computation and Language
|
String Theories involving Regular Membership Predicates: From Practice
to Theory and Back
|
Widespread use of string solvers in formal analysis of string-heavy programs
has led to a growing demand for more efficient and reliable techniques which
can be applied in this context, especially for real-world cases. Designing an
algorithm for the (generally undecidable) satisfiability problem for systems of
string constraints requires a thorough understanding of the structure of
constraints present in the targeted cases. In this paper, we investigate
benchmarks presented in the literature containing regular expression membership
predicates, extract different first order logic theories, and prove their
decidability, resp. undecidability. Notably, the most common theories in
real-world benchmarks are PSPACE-complete and directly lead to the
implementation of a more efficient algorithm to solving string constraints.
| 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.