publication_date
stringlengths 20
20
| doc_url
stringlengths 33
34
| id
stringlengths 33
34
| authors
stringlengths 5
20.1k
| update_date
stringlengths 20
20
| category_all
stringlengths 5
138
| abstract
stringlengths 123
3.09k
| category_primary
stringclasses 78
values | title
stringlengths 10
211
|
|---|---|---|---|---|---|---|---|---|
2022-12-19T05:55:58Z | http://arxiv.org/pdf/2212.09254v1 | http://arxiv.org/abs/2212.09254v1 | Bairu Hou, Jinghan Jia, Yihua Zhang, Guanhua Zhang, Yang Zhang, Sijia Liu, Shiyu Chang | 2022-12-19T05:55:58Z | cs.CL | Robustness evaluation against adversarial examples has become increasingly
important to unveil the trustworthiness of the prevailing deep models in
natural language processing (NLP). However, in contrast to the computer vision
domain where the first-order projected gradient descent (PGD) is used as the
benchmark approach to generate adversarial examples for robustness evaluation,
there lacks a principled first-order gradient-based robustness evaluation
framework in NLP. The emerging optimization challenges lie in 1) the discrete
nature of textual inputs together with the strong coupling between the
perturbation location and the actual content, and 2) the additional constraint
that the perturbed text should be fluent and achieve a low perplexity under a
language model. These challenges make the development of PGD-like NLP attacks
difficult. To bridge the gap, we propose TextGrad, a new attack generator using
gradient-driven optimization, supporting high-accuracy and high-quality
assessment of adversarial robustness in NLP. Specifically, we address the
aforementioned challenges in a unified optimization framework. And we develop
an effective convex relaxation method to co-optimize the continuously-relaxed
site selection and perturbation variables and leverage an effective sampling
method to establish an accurate mapping from the continuous optimization
variables to the discrete textual perturbations. Moreover, as a first-order
attack generation method, TextGrad can be baked into adversarial training to
further improve the robustness of NLP models. Extensive experiments are
provided to demonstrate the effectiveness of TextGrad not only in attack
generation for robustness evaluation but also in adversarial defense. | cs.CL | TextGrad: Advancing Robustness Evaluation in NLP by Gradient-Driven Optimization |
2022-12-19T05:17:33Z | http://arxiv.org/pdf/2212.09252v2 | http://arxiv.org/abs/2212.09252v2 | Sagi Shaier, Lawrence Hunter, Katharina Kann | 2022-12-20T17:52:24Z | cs.CL, cs.LG | Many dialogue systems (DSs) lack characteristics humans have, such as emotion
perception, factuality, and informativeness. Enhancing DSs with knowledge
alleviates this problem, but, as many ways of doing so exist, keeping track of
all proposed methods is difficult. Here, we present the first survey of
knowledge-enhanced DSs. We define three categories of systems - internal,
external, and hybrid - based on the knowledge they use. We survey the
motivation for enhancing DSs with knowledge, used datasets, and methods for
knowledge search, knowledge encoding, and knowledge incorporation. Finally, we
propose how to improve existing systems based on theories from linguistics and
cognitive science. | cs.CL | Mind the Knowledge Gap: A Survey of Knowledge-enhanced Dialogue Systems |
2022-12-19T05:13:52Z | http://arxiv.org/pdf/2212.09251v1 | http://arxiv.org/abs/2212.09251v1 | Ethan Perez, Sam Ringer, Kamilė Lukošiūtė, Karina Nguyen, Edwin Chen, Scott Heiner, Craig Pettit, Catherine Olsson, Sandipan Kundu, Saurav Kadavath, Andy Jones, Anna Chen, Ben Mann, Brian Israel, Bryan Seethor, Cameron McKinnon, Christopher Olah, Da Yan, Daniela Amodei, Dario Amodei, Dawn Drain, Dustin Li, Eli Tran-Johnson, Guro Khundadze, Jackson Kernion, James Landis, Jamie Kerr, Jared Mueller, Jeeyoon Hyun, Joshua Landau, Kamal Ndousse, Landon Goldberg, Liane Lovitt, Martin Lucas, Michael Sellitto, Miranda Zhang, Neerav Kingsland, Nelson Elhage, Nicholas Joseph, Noemí Mercado, Nova DasSarma, Oliver Rausch, Robin Larson, Sam McCandlish, Scott Johnston, Shauna Kravec, Sheer El Showk, Tamera Lanham, Timothy Telleen-Lawton, Tom Brown, Tom Henighan, Tristan Hume, Yuntao Bai, Zac Hatfield-Dodds, Jack Clark, Samuel R. Bowman, Amanda Askell, Roger Grosse, Danny Hernandez, Deep Ganguli, Evan Hubinger, Nicholas Schiefer, Jared Kaplan | 2022-12-19T05:13:52Z | cs.CL, cs.AI, cs.LG | As language models (LMs) scale, they develop many novel behaviors, good and
bad, exacerbating the need to evaluate how they behave. Prior work creates
evaluations with crowdwork (which is time-consuming and expensive) or existing
data sources (which are not always available). Here, we automatically generate
evaluations with LMs. We explore approaches with varying amounts of human
effort, from instructing LMs to write yes/no questions to making complex
Winogender schemas with multiple stages of LM-based generation and filtering.
Crowdworkers rate the examples as highly relevant and agree with 90-100% of
labels, sometimes more so than corresponding human-written datasets. We
generate 154 datasets and discover new cases of inverse scaling where LMs get
worse with size. Larger LMs repeat back a dialog user's preferred answer
("sycophancy") and express greater desire to pursue concerning goals like
resource acquisition and goal preservation. We also find some of the first
examples of inverse scaling in RL from Human Feedback (RLHF), where more RLHF
makes LMs worse. For example, RLHF makes LMs express stronger political views
(on gun rights and immigration) and a greater desire to avoid shut down.
Overall, LM-written evaluations are high-quality and let us quickly discover
many novel LM behaviors. | cs.CL | Discovering Language Model Behaviors with Model-Written Evaluations |
2022-12-19T05:06:00Z | http://arxiv.org/pdf/2212.09248v1 | http://arxiv.org/abs/2212.09248v1 | Pengcheng Yin, Wen-Ding Li, Kefan Xiao, Abhishek Rao, Yeming Wen, Kensen Shi, Joshua Howland, Paige Bailey, Michele Catasta, Henryk Michalewski, Alex Polozov, Charles Sutton | 2022-12-19T05:06:00Z | cs.CL, cs.SE | Computational notebooks, such as Jupyter notebooks, are interactive computing
environments that are ubiquitous among data scientists to perform data
wrangling and analytic tasks. To measure the performance of AI pair programmers
that automatically synthesize programs for those tasks given natural language
(NL) intents from users, we build ARCADE, a benchmark of 1082 code generation
problems using the pandas data analysis framework in data science notebooks.
ARCADE features multiple rounds of NL-to-code problems from the same notebook.
It requires a model to understand rich multi-modal contexts, such as existing
notebook cells and their execution states as well as previous turns of
interaction. To establish a strong baseline on this challenging task, we
develop PaChiNCo, a 62B code language model (LM) for Python computational
notebooks, which significantly outperforms public code LMs. Finally, we explore
few-shot prompting strategies to elicit better code with step-by-step
decomposition and NL explanation, showing the potential to improve the
diversity and explainability of model predictions. | cs.CL | Natural Language to Code Generation in Interactive Data Science Notebooks |
2022-12-19T04:47:49Z | http://arxiv.org/pdf/2212.09246v3 | http://arxiv.org/abs/2212.09246v3 | Chandra Bhagavatula, Jena D. Hwang, Doug Downey, Ronan Le Bras, Ximing Lu, Lianhui Qin, Keisuke Sakaguchi, Swabha Swayamdipta, Peter West, Yejin Choi | 2023-05-26T17:14:27Z | cs.CL | Commonsense capabilities of pre-trained language models dramatically improve
with scale, leading many to believe that scale is the only winning recipe. But
is it? Here, we investigate an alternative that a priori seems impossible: can
smaller language models (e.g., GPT-2) win over models that are orders of
magnitude larger and better (e.g., GPT-3), if powered with novel commonsense
distillation algorithms? The key intellectual challenge is to design a learning
algorithm that achieve a competitive level of commonsense acquisition, without
relying on the benefits of scale. In particular, we study generative models of
commonsense knowledge, focusing on the task of generating generics, statements
of commonsense facts about everyday concepts, e.g., birds can fly.
We introduce I2D2, a novel commonsense distillation framework that loosely
follows the Symbolic Knowledge Distillation of West et al. but breaks the
dependence on the extreme-scale teacher model with two innovations: (1) the
novel adaptation of NeuroLogic Decoding to enhance the generation quality of
the weak, off-the-shelf language models, and (2) self-imitation learning to
iteratively learn from the model's own enhanced commonsense acquisition
capabilities. Empirical results suggest that scale is not the only way, as
novel algorithms can be a promising alternative. Moreover, our study leads to a
new corpus of generics, Gen-A-tomic, that is the largest and highest quality
available to date. | cs.CL | I2D2: Inductive Knowledge Distillation with NeuroLogic and Self-Imitation |
2022-12-19T04:12:54Z | http://arxiv.org/pdf/2212.09235v2 | http://arxiv.org/abs/2212.09235v2 | Jiale Cheng, Sahand Sabour, Hao Sun, Zhuang Chen, Minlie Huang | 2023-05-29T06:59:14Z | cs.CL | Due to the lack of human resources for mental health support, there is an
increasing demand for employing conversational agents for support. Recent work
has demonstrated the effectiveness of dialogue models in providing emotional
support. As previous studies have demonstrated that seekers' persona is an
important factor for effective support, we investigate whether there are
benefits to modeling such information in dialogue models for support. In this
paper, our empirical analysis verifies that persona has an important impact on
emotional support. Therefore, we propose a framework for dynamically inferring
and modeling seekers' persona. We first train a model for inferring the
seeker's persona from the conversation history. Accordingly, we propose PAL, a
model that leverages persona information and, in conjunction with our
strategy-based controllable generation method, provides personalized emotional
support. Automatic and manual evaluations demonstrate that PAL achieves
state-of-the-art results, outperforming the baselines on the studied benchmark.
Our code and data are publicly available at https://github.com/chengjl19/PAL. | cs.CL | PAL: Persona-Augmented Emotional Support Conversation Generation |
2022-12-19T04:04:17Z | http://arxiv.org/pdf/2212.09233v2 | http://arxiv.org/abs/2212.09233v2 | Xianjun Yang, Kaiqiang Song, Sangwoo Cho, Xiaoyang Wang, Xiaoman Pan, Linda Petzold, Dong Yu | 2023-05-25T22:29:45Z | cs.CL | Aspect or query-based summarization has recently caught more attention, as it
can generate differentiated summaries based on users' interests. However, the
current dataset for aspect or query-based summarization either focuses on
specific domains, contains relatively small-scale instances, or includes only a
few aspect types. Such limitations hinder further explorations in this
direction. In this work, we take advantage of crowd-sourcing knowledge on
Wikipedia.org and automatically create a high-quality, large-scale open-domain
aspect-based summarization dataset named OASum, which contains more than 3.7
million instances with around 1 million different aspects on 2 million
Wikipedia pages. We provide benchmark results on OASum and demonstrate its
ability for diverse aspect-based summarization generation. To overcome the data
scarcity problem on specific domains, we also perform zero-shot, few-shot, and
fine-tuning on seven downstream datasets. Specifically, zero/few-shot and
fine-tuning results show that the model pre-trained on our corpus demonstrates
a strong aspect or query-focused generation ability compared with the backbone
model. Our dataset and pre-trained checkpoints are publicly available. | cs.CL | OASum: Large-Scale Open Domain Aspect-based Summarization |
2022-12-19T00:04:56Z | http://arxiv.org/pdf/2212.09196v3 | http://arxiv.org/abs/2212.09196v3 | Taylor Webb, Keith J. Holyoak, Hongjing Lu | 2023-08-03T03:44:47Z | cs.AI, cs.CL | The recent advent of large language models has reinvigorated debate over
whether human cognitive capacities might emerge in such generic models given
sufficient training data. Of particular interest is the ability of these models
to reason about novel problems zero-shot, without any direct training. In human
cognition, this capacity is closely tied to an ability to reason by analogy.
Here, we performed a direct comparison between human reasoners and a large
language model (the text-davinci-003 variant of GPT-3) on a range of analogical
tasks, including a non-visual matrix reasoning task based on the rule structure
of Raven's Standard Progressive Matrices. We found that GPT-3 displayed a
surprisingly strong capacity for abstract pattern induction, matching or even
surpassing human capabilities in most settings; preliminary tests of GPT-4
indicated even better performance. Our results indicate that large language
models such as GPT-3 have acquired an emergent ability to find zero-shot
solutions to a broad range of analogy problems. | cs.AI | Emergent Analogical Reasoning in Large Language Models |
2022-12-18T22:07:55Z | http://arxiv.org/pdf/2212.09180v3 | http://arxiv.org/abs/2212.09180v3 | Sarah E. Finch, James D. Finch, Jinho D. Choi | 2023-07-28T20:55:09Z | cs.CL | Despite tremendous advancements in dialogue systems, stable evaluation still
requires human judgments producing notoriously high-variance metrics due to
their inherent subjectivity. Moreover, methods and labels in dialogue
evaluation are not fully standardized, especially for open-domain chats, with a
lack of work to compare and assess the validity of those approaches. The use of
inconsistent evaluation can misinform the performance of a dialogue system,
which becomes a major hurdle to enhance it. Thus, a dimensional evaluation of
chat-oriented open-domain dialogue systems that reliably measures several
aspects of dialogue capabilities is desired. This paper presents a novel human
evaluation method to estimate the rates of many dialogue system behaviors. Our
method is used to evaluate four state-of-the-art open-domain dialogue systems
and compared with existing approaches. The analysis demonstrates that our
behavior method is more suitable than alternative Likert-style or comparative
approaches for dimensional evaluation of these systems. | cs.CL | Don't Forget Your ABC's: Evaluating the State-of-the-Art in Chat-Oriented Dialogue Systems |
2022-12-18T21:22:28Z | http://arxiv.org/pdf/2212.09171v2 | http://arxiv.org/abs/2212.09171v2 | Maxime Darrin, Pablo Piantanida, Pierre Colombo | 2023-11-01T19:06:58Z | cs.CL | Implementing effective control mechanisms to ensure the proper functioning
and security of deployed NLP models, from translation to chatbots, is
essential. A key ingredient to ensure safe system behaviour is
Out-Of-Distribution (OOD) detection, which aims to detect whether an input
sample is statistically far from the training distribution. Although OOD
detection is a widely covered topic in classification tasks, most methods rely
on hidden features output by the encoder. In this work, we focus on leveraging
soft-probabilities in a black-box framework, i.e. we can access the
soft-predictions but not the internal states of the model. Our contributions
include: (i) RAINPROOF a Relative informAItioN Projection OOD detection
framework; and (ii) a more operational evaluation setting for OOD detection.
Surprisingly, we find that OOD detection is not necessarily aligned with
task-specific measures. The OOD detector may filter out samples well processed
by the model and keep samples that are not, leading to weaker performance. Our
results show that RAINPROOF provides OOD detection methods more aligned with
task-specific performance metrics than traditional OOD detectors. | cs.CL | Rainproof: An Umbrella To Shield Text Generators From Out-Of-Distribution Data |
2022-12-18T21:11:49Z | http://arxiv.org/pdf/2212.09170v2 | http://arxiv.org/abs/2212.09170v2 | Chenghao Xiao, Yang Long, Noura Al Moubayed | 2023-05-26T20:40:57Z | cs.CL, cs.AI | Incorporating contrastive learning objectives in sentence representation
learning (SRL) has yielded significant improvements on many sentence-level NLP
tasks. However, it is not well understood why contrastive learning works for
learning sentence-level semantics. In this paper, we aim to help guide future
designs of sentence representation learning methods by taking a closer look at
contrastive SRL through the lens of isotropy, contextualization and learning
dynamics. We interpret its successes through the geometry of the representation
shifts and show that contrastive learning brings isotropy, and drives high
intra-sentence similarity: when in the same sentence, tokens converge to
similar positions in the semantic space. We also find that what we formalize as
"spurious contextualization" is mitigated for semantically meaningful tokens,
while augmented for functional ones. We find that the embedding space is
directed towards the origin during training, with more areas now better
defined. We ablate these findings by observing the learning dynamics with
different training temperatures, batch sizes and pooling methods. | cs.CL | On Isotropy, Contextualization and Learning Dynamics of Contrastive-based Sentence Representation Learning |
2022-12-18T19:27:41Z | http://arxiv.org/pdf/2212.09146v3 | http://arxiv.org/abs/2212.09146v3 | Parishad BehnamGhader, Santiago Miret, Siva Reddy | 2023-11-02T19:12:52Z | cs.CL | Augmenting pretrained language models with retrievers has shown promise in
effectively solving common NLP problems, such as language modeling and question
answering. In this paper, we evaluate the strengths and weaknesses of popular
retriever-augmented language models, namely kNN-LM, REALM, DPR + FiD,
Contriever + ATLAS, and Contriever + Flan-T5, in reasoning over retrieved
statements across different tasks. Our findings indicate that the simple
similarity metric employed by retrievers is insufficient for retrieving all the
necessary statements for reasoning. Additionally, the language models do not
exhibit strong reasoning even when provided with only the required statements.
Furthermore, when combined with imperfect retrievers, the performance of the
language models becomes even worse, e.g., Flan-T5's performance drops by 28.6%
when retrieving 5 statements using Contriever. While larger language models
improve performance, there is still a substantial room for enhancement. Our
further analysis indicates that multihop retrieve-and-read is promising for
large language models like GPT-3.5, but does not generalize to other language
models like Flan-T5-xxl. | cs.CL | Can Retriever-Augmented Language Models Reason? The Blame Game Between the Retriever and the Language Model |
2022-12-18T18:10:45Z | http://arxiv.org/pdf/2212.09140v2 | http://arxiv.org/abs/2212.09140v2 | Songlin Yang, Roger P. Levy, Yoon Kim | 2023-06-09T15:42:50Z | cs.CL | We study grammar induction with mildly context-sensitive grammars for
unsupervised discontinuous parsing. Using the probabilistic linear context-free
rewriting system (LCFRS) formalism, our approach fixes the rule structure in
advance and focuses on parameter learning with maximum likelihood. To reduce
the computational complexity of both parsing and parameter estimation, we
restrict the grammar formalism to LCFRS-2 (i.e., binary LCFRS with fan-out two)
and further discard rules that require O(n^6) time to parse, reducing inference
to O(n^5). We find that using a large number of nonterminals is beneficial and
thus make use of tensor decomposition-based rank-space dynamic programming with
an embedding-based parameterization of rule probabilities to scale up the
number of nonterminals. Experiments on German and Dutch show that our approach
is able to induce linguistically meaningful trees with continuous and
discontinuous structures | cs.CL | Unsupervised Discontinuous Constituency Parsing with Mildly Context-Sensitive Grammars |
2022-12-18T16:42:52Z | http://arxiv.org/pdf/2212.09125v1 | http://arxiv.org/abs/2212.09125v1 | Chengyue Jiang, Wenyang Hui, Yong Jiang, Xiaobin Wang, Pengjun Xie, Kewei Tu | 2022-12-18T16:42:52Z | cs.CL, cs.AI | Ultra-fine entity typing (UFET) predicts extremely free-formed types (e.g.,
president, politician) of a given entity mention (e.g., Joe Biden) in context.
State-of-the-art (SOTA) methods use the cross-encoder (CE) based architecture.
CE concatenates the mention (and its context) with each type and feeds the
pairs into a pretrained language model (PLM) to score their relevance. It
brings deeper interaction between mention and types to reach better performance
but has to perform N (type set size) forward passes to infer types of a single
mention. CE is therefore very slow in inference when the type set is large
(e.g., N = 10k for UFET). To this end, we propose to perform entity typing in a
recall-expand-filter manner. The recall and expand stages prune the large type
set and generate K (K is typically less than 256) most relevant type candidates
for each mention. At the filter stage, we use a novel model called MCCE to
concurrently encode and score these K candidates in only one forward pass to
obtain the final type prediction. We investigate different variants of MCCE and
extensive experiments show that MCCE under our paradigm reaches SOTA
performance on ultra-fine entity typing and is thousands of times faster than
the cross-encoder. We also found MCCE is very effective in fine-grained (130
types) and coarse-grained (9 types) entity typing. Our code is available at
\url{https://github.com/modelscope/AdaSeq/tree/master/examples/MCCE}. | cs.CL | Recall, Expand and Multi-Candidate Cross-Encode: Fast and Accurate Ultra-Fine Entity Typing |
2022-12-18T16:08:40Z | http://arxiv.org/pdf/2212.11126v2 | http://arxiv.org/abs/2212.11126v2 | Forrest McKee, David Noever | 2022-12-22T17:20:36Z | cs.CR, cs.CL, cs.LG | Question-and-answer formats provide a novel experimental platform for
investigating cybersecurity questions. Unlike previous chatbots, the latest
ChatGPT model from OpenAI supports an advanced understanding of complex coding
questions. The research demonstrates thirteen coding tasks that generally
qualify as stages in the MITRE ATT&CK framework, ranging from credential access
to defense evasion. With varying success, the experimental prompts generate
examples of keyloggers, logic bombs, obfuscated worms, and payment-fulfilled
ransomware. The empirical results illustrate cases that support the broad gain
of functionality, including self-replication and self-modification, evasion,
and strategic understanding of complex cybersecurity goals. One surprising
feature of ChatGPT as a language-only model centers on its ability to spawn
coding approaches that yield images that obfuscate or embed executable
programming steps or links. | cs.CR | Chatbots in a Botnet World |
2022-12-18T15:57:46Z | http://arxiv.org/pdf/2212.09114v2 | http://arxiv.org/abs/2212.09114v2 | Xingwei He, Yeyun Gong, A-Long Jin, Hang Zhang, Anlei Dong, Jian Jiao, Siu Ming Yiu, Nan Duan | 2023-10-29T09:32:07Z | cs.CL | The dual-encoder has become the de facto architecture for dense retrieval.
Typically, it computes the latent representations of the query and document
independently, thus failing to fully capture the interactions between the query
and document. To alleviate this, recent research has focused on obtaining
query-informed document representations. During training, it expands the
document with a real query, but during inference, it replaces the real query
with a generated one. This inconsistency between training and inference causes
the dense retrieval model to prioritize query information while disregarding
the document when computing the document representation. Consequently, it
performs even worse than the vanilla dense retrieval model because its
performance heavily relies on the relevance between the generated queries and
the real query.In this paper, we propose a curriculum sampling strategy that
utilizes pseudo queries during training and progressively enhances the
relevance between the generated query and the real query. By doing so, the
retrieval model learns to extend its attention from the document alone to both
the document and query, resulting in high-quality query-informed document
representations. Experimental results on both in-domain and out-of-domain
datasets demonstrate that our approach outperforms previous dense retrieval
models. | cs.CL | CAPSTONE: Curriculum Sampling for Dense Retrieval with Document Expansion |
2022-12-18T15:10:05Z | http://arxiv.org/pdf/2212.09104v1 | http://arxiv.org/abs/2212.09104v1 | Sayan Ghosh, Rakesh R Menon, Shashank Srivastava | 2022-12-18T15:10:05Z | cs.CL | A hallmark of human intelligence is the ability to learn new concepts purely
from language. Several recent approaches have explored training machine
learning models via natural language supervision. However, these approaches
fall short in leveraging linguistic quantifiers (such as 'always' or 'rarely')
and mimicking humans in compositionally learning complex tasks. Here, we
present LaSQuE, a method that can learn zero-shot classifiers from language
explanations by using three new strategies - (1) modeling the semantics of
linguistic quantifiers in explanations (including exploiting ordinal strength
relationships, such as 'always' > 'likely'), (2) aggregating information from
multiple explanations using an attention-based mechanism, and (3) model
training via curriculum learning. With these strategies, LaSQuE outperforms
prior work, showing an absolute gain of up to 7% in generalizing to unseen
real-world classification tasks. | cs.CL | LaSQuE: Improved Zero-Shot Classification from Explanations Through Quantifier Modeling and Curriculum Learning |
2022-12-18T14:41:13Z | http://arxiv.org/pdf/2212.09097v2 | http://arxiv.org/abs/2212.09097v2 | Yuanchi Zhang, Peng Li, Maosong Sun, Yang Liu | 2023-06-12T12:00:20Z | cs.CL | While many parallel corpora are not publicly accessible for data copyright,
data privacy and competitive differentiation reasons, trained translation
models are increasingly available on open platforms. In this work, we propose a
method called continual knowledge distillation to take advantage of existing
translation models to improve one model of interest. The basic idea is to
sequentially transfer knowledge from each trained model to the distilled model.
Extensive experiments on Chinese-English and German-English datasets show that
our method achieves significant and consistent improvements over strong
baselines under both homogeneous and heterogeneous trained model settings and
is robust to malicious models. | cs.CL | Continual Knowledge Distillation for Neural Machine Translation |
2022-12-18T14:36:07Z | http://arxiv.org/pdf/2212.09095v2 | http://arxiv.org/abs/2212.09095v2 | Hritik Bansal, Karthik Gopalakrishnan, Saket Dingliwal, Sravan Bodapati, Katrin Kirchhoff, Dan Roth | 2023-08-16T09:09:53Z | cs.CL, cs.AI | Language models have been shown to perform better with an increase in scale
on a wide variety of tasks via the in-context learning paradigm. In this paper,
we investigate the hypothesis that the ability of a large language model to
in-context learn-perform a task is not uniformly spread across all of its
underlying components. Using a 66 billion parameter language model (OPT-66B)
across a diverse set of 14 downstream tasks, we find this is indeed the case:
$\sim$70% of attention heads and $\sim$20% of feed forward networks can be
removed with minimal decline in task performance. We find substantial overlap
in the set of attention heads (un)important for in-context learning across
tasks and number of in-context examples. We also address our hypothesis through
a task-agnostic lens, finding that a small set of attention heads in OPT-66B
score highly on their ability to perform primitive induction operations
associated with in-context learning, namely, prefix matching and copying. These
induction heads overlap with task-specific important heads, reinforcing
arguments by Olsson et al. (arXiv:2209.11895) regarding induction head
generality to more sophisticated behaviors associated with in-context learning.
Overall, our study provides several insights that indicate large language
models may be under-trained for in-context learning and opens up questions on
how to pre-train language models to more effectively perform in-context
learning. | cs.CL | Rethinking the Role of Scale for In-Context Learning: An Interpretability-based Case Study at 66 Billion Scale |
2022-12-18T13:36:07Z | http://arxiv.org/pdf/2212.09086v4 | http://arxiv.org/abs/2212.09086v4 | Yongkang Liu, Shi Feng, Daling Wang, Yifei Zhang, Hinrich Schütze | 2023-05-16T11:29:24Z | cs.CL | We investigate response generation for multi-turn dialogue in
generative-based chatbots. Existing generative models based on RNNs (Recurrent
Neural Networks) usually employ the last hidden state to summarize the
sequences, which makes models unable to capture the subtle variability observed
in different dialogues and cannot distinguish the differences between dialogues
that are similar in composition. In this paper, we propose a Pseudo-Variational
Gated Recurrent Unit (PVGRU) component without posterior knowledge through
introducing a recurrent summarizing variable into the GRU, which can aggregate
the accumulated distribution variations of subsequences. PVGRU can perceive the
subtle semantic variability through summarizing variables that are optimized by
the devised distribution consistency and reconstruction objectives. In
addition, we build a Pseudo-Variational Hierarchical Dialogue (PVHD) model
based on PVGRU. Experimental results demonstrate that PVGRU can broadly improve
the diversity and relevance of responses on two benchmark datasets. | cs.CL | PVGRU: Generating Diverse and Relevant Dialogue Responses via Pseudo-Variational Mechanism |
2022-12-18T12:54:45Z | http://arxiv.org/pdf/2212.09080v1 | http://arxiv.org/abs/2212.09080v1 | Akila Peiris, Nisansa de Silva | 2022-12-18T12:54:45Z | cs.CL, cs.LG | This paper introduces the Forgotten Realms Wiki (FRW) data set and domain
specific natural language generation using FRW along with related analyses.
Forgotten Realms is the de-facto default setting of the popular open ended
tabletop fantasy role playing game, Dungeons & Dragons. The data set was
extracted from the Forgotten Realms Fandom wiki consisting of more than over
45,200 articles. The FRW data set is constituted of 11 sub-data sets in a
number of formats: raw plain text, plain text annotated by article title,
directed link graphs, wiki info-boxes annotated by the wiki article title,
Poincar\'e embedding of first link graph, multiple Word2Vec and Doc2Vec models
of the corpus. This is the first data set of this size for the Dungeons &
Dragons domain. We then present a pairwise similarity comparison benchmark
which utilizes similarity measures. In addition, we perform D&D domain specific
natural language generation using the corpus and evaluate the named entity
classification with respect to the lore of Forgotten Realms. | cs.CL | Synthesis and Evaluation of a Domain-specific Large Data Set for Dungeons & Dragons |
2022-12-18T12:03:53Z | http://arxiv.org/pdf/2212.09072v1 | http://arxiv.org/abs/2212.09072v1 | Haolan Zhan, Yufei Wang, Tao Feng, Yuncheng Hua, Suraj Sharma, Zhuang Li, Lizhen Qu, Gholamreza Haffari | 2022-12-18T12:03:53Z | cs.CL | Negotiation is one of the crucial abilities in human communication, and there
has been a resurgent research interest in negotiation dialogue systems
recently, which goal is to empower intelligent agents with such ability that
can efficiently help humans resolve conflicts or reach beneficial agreements.
Although there have been many explorations in negotiation dialogue systems, a
systematic review of this task has to date remained notably absent. To this
end, we aim to fill this gap by reviewing contemporary studies in the emerging
field of negotiation dialogue systems, covering benchmarks, evaluations, and
methodologies. Furthermore, we also discuss potential future directions,
including multi-modal, multi-party, and cross-cultural negotiation scenarios.
Our goal is to provide the community with a systematic overview of negotiation
dialogue systems and to inspire future research. | cs.CL | Let's Negotiate! A Survey of Negotiation Dialogue Systems |
2022-12-18T10:41:55Z | http://arxiv.org/pdf/2212.09058v1 | http://arxiv.org/abs/2212.09058v1 | Sanyuan Chen, Yu Wu, Chengyi Wang, Shujie Liu, Daniel Tompkins, Zhuo Chen, Furu Wei | 2022-12-18T10:41:55Z | eess.AS, cs.AI, cs.CL, cs.LG, cs.SD | The massive growth of self-supervised learning (SSL) has been witnessed in
language, vision, speech, and audio domains over the past few years. While
discrete label prediction is widely adopted for other modalities, the
state-of-the-art audio SSL models still employ reconstruction loss for
pre-training. Compared with reconstruction loss, semantic-rich discrete label
prediction encourages the SSL model to abstract the high-level audio semantics
and discard the redundant details as in human perception. However, a
semantic-rich acoustic tokenizer for general audio pre-training is usually not
straightforward to obtain, due to the continuous property of audio and
unavailable phoneme sequences like speech. To tackle this challenge, we propose
BEATs, an iterative audio pre-training framework to learn Bidirectional Encoder
representation from Audio Transformers, where an acoustic tokenizer and an
audio SSL model are optimized by iterations. In the first iteration, we use
random projection as the acoustic tokenizer to train an audio SSL model in a
mask and label prediction manner. Then, we train an acoustic tokenizer for the
next iteration by distilling the semantic knowledge from the pre-trained or
fine-tuned audio SSL model. The iteration is repeated with the hope of mutual
promotion of the acoustic tokenizer and audio SSL model. The experimental
results demonstrate our acoustic tokenizers can generate discrete labels with
rich audio semantics and our audio SSL models achieve state-of-the-art results
across various audio classification benchmarks, even outperforming previous
models that use more training data and model parameters significantly.
Specifically, we set a new state-of-the-art mAP 50.6% on AudioSet-2M for
audio-only models without using any external data, and 98.1% accuracy on
ESC-50. The code and pre-trained models are available at https://aka.ms/beats. | eess.AS | BEATs: Audio Pre-Training with Acoustic Tokenizers |
2022-12-18T10:18:15Z | http://arxiv.org/pdf/2212.09056v1 | http://arxiv.org/abs/2212.09056v1 | Rishav Hada, Amir Ebrahimi Fard, Sarah Shugars, Federico Bianchi, Patricia Rossini, Dirk Hovy, Rebekah Tromble, Nava Tintarev | 2022-12-18T10:18:15Z | cs.CL | Increasingly taking place in online spaces, modern political conversations
are typically perceived to be unproductively affirming -- siloed in so called
``echo chambers'' of exclusively like-minded discussants. Yet, to date we lack
sufficient means to measure viewpoint diversity in conversations. To this end,
in this paper, we operationalize two viewpoint metrics proposed for recommender
systems and adapt them to the context of social media conversations. This is
the first study to apply these two metrics (Representation and Fragmentation)
to real world data and to consider the implications for online conversations
specifically. We apply these measures to two topics -- daylight savings time
(DST), which serves as a control, and the more politically polarized topic of
immigration. We find that the diversity scores for both Fragmentation and
Representation are lower for immigration than for DST. Further, we find that
while pro-immigrant views receive consistent pushback on the platform,
anti-immigrant views largely operate within echo chambers. We observe less
severe yet similar patterns for DST. Taken together, Representation and
Fragmentation paint a meaningful and important new picture of viewpoint
diversity. | cs.CL | Beyond Digital "Echo Chambers": The Role of Viewpoint Diversity in Political Discussion |
2022-12-18T09:57:40Z | http://arxiv.org/pdf/2212.09052v1 | http://arxiv.org/abs/2212.09052v1 | Yuncong Li, Fang Wang, Sheng-Hua Zhong | 2022-12-18T09:57:40Z | cs.CL | Aspect sentiment triplet extraction (ASTE) aims to extract aspect term,
sentiment and opinion term triplets from sentences. Since the initial datasets
used to evaluate models on ASTE had flaws, several studies later corrected the
initial datasets and released new versions of the datasets independently. As a
result, different studies select different versions of datasets to evaluate
their methods, which makes ASTE-related works hard to follow. In this paper, we
analyze the relation between different versions of datasets and suggest that
the entire-space version should be used for ASTE. Besides the sentences
containing triplets and the triplets in the sentences, the entire-space version
additionally includes the sentences without triplets and the aspect terms which
do not belong to any triplets. Hence, the entire-space version is consistent
with real-world scenarios and evaluating models on the entire-space version can
better reflect the models' performance in real-world scenarios. In addition,
experimental results show that evaluating models on non-entire-space datasets
inflates the performance of existing models and models trained on the
entire-space version can obtain better performance. | cs.CL | A Better Choice: Entire-space Datasets for Aspect Sentiment Triplet Extraction |
2022-12-18T09:37:20Z | http://arxiv.org/pdf/2212.09045v1 | http://arxiv.org/abs/2212.09045v1 | Sebastin Santy, Prasanta Bhattacharya, Rishabh Mehrotra | 2022-12-18T09:37:20Z | cs.CL, cs.CY, cs.HC, cs.SI | With the steady emergence of community question answering (CQA) platforms
like Quora, StackExchange, and WikiHow, users now have an unprecedented access
to information on various kind of queries and tasks. Moreover, the rapid
proliferation and localization of these platforms spanning geographic and
linguistic boundaries offer a unique opportunity to study the task requirements
and preferences of users in different socio-linguistic groups. In this study,
we implement an entity-embedding model trained on a large longitudinal dataset
of multi-lingual and task-oriented question-answer pairs to uncover and
quantify the (i) prevalence and distribution of various online tasks across
linguistic communities, and (ii) emerging and receding trends in task
popularity over time in these communities. Our results show that there exists
substantial variance in task preference as well as popularity trends across
linguistic communities on the platform. Findings from this study will help Q&A
platforms better curate and personalize content for non-English users, while
also offering valuable insights to businesses looking to target non-English
speaking communities online. | cs.CL | Task Preferences across Languages on Community Question Answering Platforms |
2022-12-18T07:36:35Z | http://arxiv.org/pdf/2212.09028v1 | http://arxiv.org/abs/2212.09028v1 | Yu Wang, Hongxia Jin | 2022-12-18T07:36:35Z | cs.CL | The target of a coreference resolution system is to cluster all mentions that
refer to the same entity in a given context. All coreference resolution systems
need to solve two subtasks; one task is to detect all of the potential
mentions, and the other is to learn the linking of an antecedent for each
possible mention. In this paper, we propose a reinforcement learning
actor-critic-based neural coreference resolution system, which can achieve both
mention detection and mention clustering by leveraging an actor-critic deep
reinforcement learning technique and a joint training algorithm. We experiment
on the BERT model to generate different input span representations. Our model
with the BERT span representation achieves the state-of-the-art performance
among the models on the CoNLL-2012 Shared Task English Test Set. | cs.CL | Neural Coreference Resolution based on Reinforcement Learning |
2022-12-18T03:53:44Z | http://arxiv.org/pdf/2212.08999v1 | http://arxiv.org/abs/2212.08999v1 | Shabnam Behzad, Amir Zeldes, Nathan Schneider | 2022-12-18T03:53:44Z | cs.CL | In this paper, we present strong baselines for the task of Feedback Comment
Generation for Writing Learning. Given a sentence and an error span, the task
is to generate a feedback comment explaining the error. Sentences and feedback
comments are both in English. We experiment with LLMs and also create multiple
pseudo datasets for the task, investigating how it affects the performance of
our system. We present our results for the task along with extensive analysis
of the generated comments with the aim of aiding future studies in feedback
comment generation for English language learners. | cs.CL | Sentence-level Feedback Generation for English Language Learners: Does Data Augmentation Help? |
2022-12-18T03:17:47Z | http://arxiv.org/pdf/2212.08995v1 | http://arxiv.org/abs/2212.08995v1 | Amira Yousif, James Buckley | 2022-12-18T03:17:47Z | cs.CL, cs.LG | Fake review identification is an important topic and has gained the interest
of experts all around the world. Identifying fake reviews is challenging for
researchers, and there are several primary challenges to fake review detection.
We propose developing an initial research paper for investigating fake reviews
by using sentiment analysis. Ten research papers are identified that show fake
reviews, and they discuss currently available solutions for predicting or
detecting fake reviews. They also show the distribution of fake and truthful
reviews through the analysis of sentiment. We summarize and compare previous
studies related to fake reviews. We highlight the most significant challenges
in the sentiment evaluation process and demonstrate that there is a significant
impact on sentiment scores used to identify fake feedback. | cs.CL | Impact of Sentiment Analysis in Fake Review Detection |
2022-12-18T02:26:50Z | http://arxiv.org/pdf/2212.08992v1 | http://arxiv.org/abs/2212.08992v1 | Chen Zhang, Luis Fernando D'Haro, Qiquan Zhang, Thomas Friedrichs, Haizhou Li | 2022-12-18T02:26:50Z | cs.CL | Chatbots are expected to be knowledgeable across multiple domains, e.g. for
daily chit-chat, exchange of information, and grounding in emotional
situations. To effectively measure the quality of such conversational agents, a
model-based automatic dialogue evaluation metric (ADEM) is expected to perform
well across multiple domains. Despite significant progress, an ADEM that works
well in one domain does not necessarily generalize to another. This calls for a
dedicated network architecture for domain generalization. To tackle the
multi-domain dialogue evaluation task, we propose a Panel of Experts (PoE), a
multitask network that consists of a shared transformer encoder and a
collection of lightweight adapters. The shared encoder captures the general
knowledge of dialogues across domains, while each adapter specializes in one
specific domain and serves as a domain expert. To validate the idea, we
construct a high-quality multi-domain dialogue dataset leveraging data
augmentation and pseudo-labeling. The PoE network is comprehensively assessed
on 16 dialogue evaluation datasets spanning a wide range of dialogue domains.
It achieves state-of-the-art performance in terms of mean Spearman correlation
over all the evaluation datasets. It exhibits better zero-shot generalization
than existing state-of-the-art ADEMs and the ability to easily adapt to new
domains with few-shot transfer learning. | cs.CL | PoE: a Panel of Experts for Generalized Automatic Dialogue Assessment |
2022-12-18T01:59:49Z | http://arxiv.org/pdf/2212.08987v1 | http://arxiv.org/abs/2212.08987v1 | Yu Wang, Hongxia Jin | 2022-12-18T01:59:49Z | cs.CL, cs.AI | Most recent semantic frame parsing systems for spoken language understanding
(SLU) are designed based on recurrent neural networks. These systems display
decent performance on benchmark SLU datasets such as ATIS or SNIPS, which
contain short utterances with relatively simple patterns. However, the current
semantic frame parsing models lack a mechanism to handle out-of-distribution
(\emph{OOD}) patterns and out-of-vocabulary (\emph{OOV}) tokens. In this paper,
we introduce a robust semantic frame parsing pipeline that can handle both
\emph{OOD} patterns and \emph{OOV} tokens in conjunction with a new complex
Twitter dataset that contains long tweets with more \emph{OOD} patterns and
\emph{OOV} tokens. The new pipeline demonstrates much better results in
comparison to state-of-the-art baseline SLU models on both the SNIPS dataset
and the new Twitter dataset (Our new Twitter dataset can be downloaded from
https://1drv.ms/u/s!AroHb-W6_OAlavK4begsDsMALfE?e=c8f2XX ). Finally, we also
build an E2E application to demo the feasibility of our algorithm and show why
it is useful in real application. | cs.CL | A Robust Semantic Frame Parsing Pipeline on a New Complex Twitter Dataset |
2022-12-18T01:57:30Z | http://arxiv.org/pdf/2212.08986v2 | http://arxiv.org/abs/2212.08986v2 | Ajay Patel, Nicholas Andrews, Chris Callison-Burch | 2023-08-23T16:31:34Z | cs.CL | Authorship style transfer involves altering text to match the style of a
target author whilst preserving the original meaning. Existing unsupervised
approaches like STRAP have largely focused on style transfer to target authors
with many examples of their writing style in books, speeches, or other
published works. This high-resource training data requirement (often greater
than 100,000 words) makes these approaches primarily useful for style transfer
to published authors, politicians, or other well-known figures and authorship
styles, while style transfer to non-famous authors has not been well-studied.
We introduce the \textit{low-resource authorship style transfer} task, a more
challenging class of authorship style transfer where only a limited amount of
text in the target author's style may exist. In our experiments, we
specifically choose source and target authors from Reddit and style transfer
their Reddit posts, limiting ourselves to just 16 posts (on average ~500 words)
of the target author's style. Style transfer accuracy is typically measured by
how often a classifier or human judge will classify an output as written by the
target author. Recent authorship representations models excel at authorship
identification even with just a few writing samples, making automatic
evaluation of this task possible for the first time through evaluation metrics
we propose. Our results establish an in-context learning technique we develop
as the strongest baseline, though we find current approaches do not yet achieve
mastery of this challenging task. We release our data and implementations to
encourage further investigation. | cs.CL | Low-Resource Authorship Style Transfer: Can Non-Famous Authors Be Imitated? |
2022-12-18T00:11:06Z | http://arxiv.org/pdf/2212.08979v1 | http://arxiv.org/abs/2212.08979v1 | Koustuv Sinha, Jon Gauthier, Aaron Mueller, Kanishka Misra, Keren Fuentes, Roger Levy, Adina Williams | 2022-12-18T00:11:06Z | cs.CL, cs.LG | Targeted syntactic evaluations of language models ask whether models show
stable preferences for syntactically acceptable content over minimal-pair
unacceptable inputs. Most targeted syntactic evaluation datasets ask models to
make these judgements with just a single context-free sentence as input. This
does not match language models' training regime, in which input sentences are
always highly contextualized by the surrounding corpus. This mismatch raises an
important question: how robust are models' syntactic judgements in different
contexts? In this paper, we investigate the stability of language models'
performance on targeted syntactic evaluations as we vary properties of the
input context: the length of the context, the types of syntactic phenomena it
contains, and whether or not there are violations of grammaticality. We find
that model judgements are generally robust when placed in randomly sampled
linguistic contexts. However, they are substantially unstable for contexts
containing syntactic structures matching those in the critical test content.
Among all tested models (GPT-2 and five variants of OPT), we significantly
improve models' judgements by providing contexts with matching syntactic
structures, and conversely significantly worsen them using unacceptable
contexts with matching but violated syntactic structures. This effect is
amplified by the length of the context, except for unrelated inputs. We show
that these changes in model performance are not explainable by simple features
matching the context and the test inputs, such as lexical overlap and
dependency overlap. This sensitivity to highly specific syntactic features of
the context can only be explained by the models' implicit in-context learning
abilities. | cs.CL | Language model acceptability judgements are not always robust to context |
2022-12-17T20:45:59Z | http://arxiv.org/pdf/2212.08950v1 | http://arxiv.org/abs/2212.08950v1 | Iman Hosseini, Brendan Dolan-Gavitt | 2022-12-17T20:45:59Z | cs.CR, cs.CL, cs.PL | The problem of reversing the compilation process, decompilation, is an
important tool in reverse engineering of computer software. Recently,
researchers have proposed using techniques from neural machine translation to
automate the process in decompilation. Although such techniques hold the
promise of targeting a wider range of source and assembly languages, to date
they have primarily targeted C code. In this paper we argue that existing
neural decompilers have achieved higher accuracy at the cost of requiring
language-specific domain knowledge such as tokenizers and parsers to build an
abstract syntax tree (AST) for the source language, which increases the
overhead of supporting new languages. We explore a different tradeoff that, to
the extent possible, treats the assembly and source languages as plain text,
and show that this allows us to build a decompiler that is easily retargetable
to new languages. We evaluate our prototype decompiler, Beyond The C (BTC), on
Go, Fortran, OCaml, and C, and examine the impact of parameters such as
tokenization and training data selection on the quality of decompilation,
finding that it achieves comparable decompilation results to prior work in
neural decompilation with significantly less domain knowledge. We will release
our training data, trained decompilation models, and code to help encourage
future research into language-agnostic decompilation. | cs.CR | Beyond the C: Retargetable Decompilation using Neural Machine Translation |
2022-12-17T20:36:17Z | http://arxiv.org/pdf/2212.08946v1 | http://arxiv.org/abs/2212.08946v1 | Shaomu Tan, Denis Paperno | 2022-12-17T20:36:17Z | cs.CL | In many real-world scenarios, the absence of external knowledge source like
Wikipedia restricts question answering systems to rely on latent internal
knowledge in limited dialogue data. In addition, humans often seek answers by
asking several questions for more comprehensive information. As the dialog
becomes more extensive, machines are challenged to refer to previous
conversation rounds to answer questions. In this work, we propose to leverage
latent knowledge in existing conversation logs via a neural Retrieval-Reading
system, enhanced with a TFIDF-based text summarizer refining lengthy
conversational history to alleviate the long context issue. Our experiments
show that our Retrieval-Reading system can exploit retrieved background
knowledge to generate significantly better answers. The results also indicate
that our context summarizer significantly helps both the retriever and the
reader by introducing more concise and less noisy contextual information. | cs.CL | Towards leveraging latent knowledge and Dialogue context for real-world conversational question answering |
2022-12-17T18:45:23Z | http://arxiv.org/pdf/2212.08929v2 | http://arxiv.org/abs/2212.08929v2 | Zixia Jia, Zhaohui Yan, Wenjuan Han, Zilong Zheng, Kewei Tu | 2023-05-28T09:48:24Z | cs.CL | Prior works on joint Information Extraction (IE) typically model instance
(e.g., event triggers, entities, roles, relations) interactions by
representation enhancement, type dependencies scoring, or global decoding. We
find that the previous models generally consider binary type dependency scoring
of a pair of instances, and leverage local search such as beam search to
approximate global solutions. To better integrate cross-instance interactions,
in this work, we introduce a joint IE framework (CRFIE) that formulates joint
IE as a high-order Conditional Random Field. Specifically, we design binary
factors and ternary factors to directly model interactions between not only a
pair of instances but also triplets. Then, these factors are utilized to
jointly predict labels of all instances. To address the intractability problem
of exact high-order inference, we incorporate a high-order neural decoder that
is unfolded from a mean-field variational inference method, which achieves
consistent learning and inference. The experimental results show that our
approach achieves consistent improvements on three IE tasks compared with our
baseline and prior work. | cs.CL | Modeling Instance Interactions for Joint Information Extraction with Neural High-Order Conditional Random Field |
2022-12-17T18:22:20Z | http://arxiv.org/pdf/2212.08926v1 | http://arxiv.org/abs/2212.08926v1 | Lior Vassertail, Omer Levy | 2022-12-17T18:22:20Z | cs.CL, cs.AI, cs.LG | Reranking methods in machine translation aim to close the gap between common
evaluation metrics (e.g. BLEU) and maximum likelihood learning and decoding
algorithms. Prior works address this challenge by training models to rerank
beam search candidates according to their predicted BLEU scores, building upon
large models pretrained on massive monolingual corpora -- a privilege that was
never made available to the baseline translation model. In this work, we
examine a simple approach for training rerankers to predict translation
candidates' BLEU scores without introducing additional data or parameters. Our
approach can be used as a clean baseline, decoupled from external factors, for
future research in this area. | cs.CL | A Simple Baseline for Beam Search Reranking |
2022-12-17T16:30:27Z | http://arxiv.org/pdf/2212.08913v2 | http://arxiv.org/abs/2212.08913v2 | Gabriella Skitalinskaya, Maximilian Spliethöver, Henning Wachsmuth | 2023-09-07T09:01:20Z | cs.CL | An optimal delivery of arguments is key to persuasion in any debate, both for
humans and for AI systems. This requires the use of clear and fluent claims
relevant to the given debate. Prior work has studied the automatic assessment
of argument quality extensively. Yet, no approach actually improves the quality
so far. To fill this gap, this paper proposes the task of claim optimization:
to rewrite argumentative claims in order to optimize their delivery. As
multiple types of optimization are possible, we approach this task by first
generating a diverse set of candidate claims using a large language model, such
as BART, taking into account contextual information. Then, the best candidate
is selected using various quality metrics. In automatic and human evaluation on
an English-language corpus, our quality-based candidate selection outperforms
several baselines, improving 60% of all claims (worsening 16% only). Follow-up
analyses reveal that, beyond copy editing, our approach often specifies claims
with details, whereas it adds less evidence than humans do. Moreover, its
capabilities generalize well to other domains, such as instructional texts. | cs.CL | Claim Optimization in Computational Argumentation |
2022-12-17T16:14:30Z | http://arxiv.org/pdf/2212.08911v1 | http://arxiv.org/abs/2212.08911v1 | Xingshan Zeng, Liangyou Li, Qun Liu | 2022-12-17T16:14:30Z | cs.CL, cs.SD, eess.AS | To alleviate the data scarcity problem in End-to-end speech translation (ST),
pre-training on data for speech recognition and machine translation is
considered as an important technique. However, the modality gap between speech
and text prevents the ST model from efficiently inheriting knowledge from the
pre-trained models. In this work, we propose AdaTranS for end-to-end ST. It
adapts the speech features with a new shrinking mechanism to mitigate the
length mismatch between speech and text features by predicting word boundaries.
Experiments on the MUST-C dataset demonstrate that AdaTranS achieves better
performance than the other shrinking-based methods, with higher inference speed
and lower memory usage. Further experiments also show that AdaTranS can be
equipped with additional alignment losses to further improve performance. | cs.CL | AdaTranS: Adapting with Boundary-based Shrinking for End-to-End Speech Translation |
2022-12-17T16:05:50Z | http://arxiv.org/pdf/2212.08909v2 | http://arxiv.org/abs/2212.08909v2 | Yifan Wang, Zewei Sun, Shanbo Cheng, Weiguo Zheng, Mingxuan Wang | 2023-05-29T02:59:07Z | cs.CL | Controlling styles in neural machine translation (NMT) has attracted wide
attention, as it is crucial for enhancing user experience. Earlier studies on
this topic typically concentrate on regulating the level of formality and
achieve some progress in this area. However, they still encounter two major
challenges. The first is the difficulty in style evaluation. The style
comprises various aspects such as lexis, syntax, and others that provide
abundant information. Nevertheless, only formality has been thoroughly
investigated. The second challenge involves excessive dependence on incremental
adjustments, particularly when new styles are necessary. To address both
challenges, this paper presents a new benchmark and approach. A multiway
stylized machine translation (MSMT) benchmark is introduced, incorporating
diverse categories of styles across four linguistic domains. Then, we propose a
method named style activation prompt (StyleAP) by retrieving prompts from
stylized monolingual corpus, which does not require extra fine-tuning.
Experiments show that StyleAP could effectively control the style of
translation and achieve remarkable performance. | cs.CL | Controlling Styles in Neural Machine Translation with Activation Prompt |
2022-12-17T15:32:00Z | http://arxiv.org/pdf/2212.08902v2 | http://arxiv.org/abs/2212.08902v2 | Bing Wang, Yan Gao, Zhoujun Li, Jian-Guang Lou | 2023-05-19T14:27:31Z | cs.CL | The task of text-to-SQL aims to convert a natural language question into its
corresponding SQL query within the context of relational tables. Existing
text-to-SQL parsers generate a "plausible" SQL query for an arbitrary user
question, thereby failing to correctly handle problematic user questions. To
formalize this problem, we conduct a preliminary study on the observed
ambiguous and unanswerable cases in text-to-SQL and summarize them into 6
feature categories. Correspondingly, we identify the causes behind each
category and propose requirements for handling ambiguous and unanswerable
questions. Following this study, we propose a simple yet effective
counterfactual example generation approach that automatically produces
ambiguous and unanswerable text-to-SQL examples. Furthermore, we propose a
weakly supervised DTE (Detecting-Then-Explaining) model for error detection,
localization, and explanation. Experimental results show that our model
achieves the best result on both real-world examples and generated examples
compared with various baselines. We release our data and code at:
\href{https://github.com/wbbeyourself/DTE}{https://github.com/wbbeyourself/DTE}. | cs.CL | Know What I don't Know: Handling Ambiguous and Unanswerable Questions for Text-to-SQL |
2022-12-17T15:20:18Z | http://arxiv.org/pdf/2212.08897v2 | http://arxiv.org/abs/2212.08897v2 | Piotr Rybak, Piotr Przybyła, Maciej Ogrodniczuk | 2024-02-22T13:24:49Z | cs.CL | Recently proposed systems for open-domain question answering (OpenQA) require
large amounts of training data to achieve state-of-the-art performance.
However, data annotation is known to be time-consuming and therefore expensive
to acquire. As a result, the appropriate datasets are available only for a
handful of languages (mainly English and Chinese). In this work, we introduce
and publicly release PolQA, the first Polish dataset for OpenQA. It consists of
7,000 questions, 87,525 manually labeled evidence passages, and a corpus of
over 7,097,322 candidate passages. Each question is classified according to its
formulation, type, as well as entity type of the answer. This resource allows
us to evaluate the impact of different annotation choices on the performance of
the QA system and propose an efficient annotation strategy that increases the
passage retrieval accuracy@10 by 10.55 p.p. while reducing the annotation cost
by 82%. | cs.CL | PolQA: Polish Question Answering Dataset |
2022-12-17T14:57:52Z | http://arxiv.org/pdf/2212.08888v1 | http://arxiv.org/abs/2212.08888v1 | Chenyang Lyu, Linyi Yang, Yue Zhang, Yvette Graham, Jennifer Foster | 2022-12-17T14:57:52Z | cs.CL | User and product information associated with a review is useful for sentiment
polarity prediction. Typical approaches incorporating such information focus on
modeling users and products as implicitly learned representation vectors. Most
do not exploit the potential of historical reviews, or those that currently do
require unnecessary modifications to model architecture or do not make full use
of user/product associations. The contribution of this work is twofold: i) a
method to explicitly employ historical reviews belonging to the same
user/product to initialize representations, and ii) efficient incorporation of
textual associations between users and products via a user-product
cross-context module. Experiments on IMDb, Yelp-2013 and Yelp-2014 benchmarks
show that our approach substantially outperforms previous state-of-the-art.
Since we employ BERT-base as the encoder, we additionally provide experiments
in which our approach performs well with Span-BERT and Longformer. Furthermore,
experiments where the reviews of each user/product in the training data are
downsampled demonstrate the effectiveness of our approach under a low-resource
setting. | cs.CL | Exploiting Rich Textual User-Product Context for Improving Sentiment Analysis |
2022-12-17T13:08:39Z | http://arxiv.org/pdf/2212.08864v1 | http://arxiv.org/abs/2212.08864v1 | Alapan Kuila, Sudeshna Sarkar | 2022-12-17T13:08:39Z | cs.CL, cs.AI, cs.CY | The coverage of different stakeholders mentioned in the news articles
significantly impacts the slant or polarity detection of the concerned news
publishers. For instance, the pro-government media outlets would give more
coverage to the government stakeholders to increase their accessibility to the
news audiences. In contrast, the anti-government news agencies would focus more
on the views of the opponent stakeholders to inform the readers about the
shortcomings of government policies. In this paper, we address the problem of
stakeholder extraction from news articles and thereby determine the inherent
bias present in news reporting. Identifying potential stakeholders in
multi-topic news scenarios is challenging because each news topic has different
stakeholders. The research presented in this paper utilizes both contextual
information and external knowledge to identify the topic-specific stakeholders
from news articles. We also apply a sequential incremental clustering algorithm
to group the entities with similar stakeholder types. We carried out all our
experiments on news articles on four Indian government policies published by
numerous national and international news agencies. We also further generalize
our system, and the experimental results show that the proposed model can be
extended to other news topics. | cs.CL | 'If you build they will come': Automatic Identification of News-Stakeholders to detect Party Preference in News Coverage |
2022-12-17T11:56:21Z | http://arxiv.org/pdf/2212.08853v2 | http://arxiv.org/abs/2212.08853v2 | Hongyi Yuan, Zheng Yuan, Chuanqi Tan, Fei Huang, Songfang Huang | 2023-05-11T07:10:32Z | cs.CL | Language models with the Transformers structure have shown great performance
in natural language processing. However, there still poses problems when
fine-tuning pre-trained language models on downstream tasks, such as
over-fitting or representation collapse. In this work, we propose HyPe, a
simple yet effective fine-tuning technique to alleviate such problems by
perturbing hidden representations of Transformers layers. Unlike previous works
that only add noise to inputs or parameters, we argue that the hidden
representations of Transformers layers convey more diverse and meaningful
language information. Therefore, making the Transformers layers more robust to
hidden representation perturbations can further benefit the fine-tuning of PLMs
en bloc. We conduct extensive experiments and analyses on GLUE and other
natural language inference datasets. Results demonstrate that HyPe outperforms
vanilla fine-tuning and enhances generalization of hidden representations from
different layers. In addition, HyPe acquires negligible computational
overheads, and is better than and compatible with previous state-of-the-art
fine-tuning techniques. | cs.CL | HyPe: Better Pre-trained Language Model Fine-tuning with Hidden Representation Perturbation |
2022-12-17T10:43:25Z | http://arxiv.org/pdf/2212.08841v2 | http://arxiv.org/abs/2212.08841v2 | Rui Meng, Ye Liu, Semih Yavuz, Divyansh Agarwal, Lifu Tu, Ning Yu, Jianguo Zhang, Meghana Bhat, Yingbo Zhou | 2023-03-07T20:51:26Z | cs.CL, cs.IR | Dense retrievers have made significant strides in text retrieval and
open-domain question answering, even though most achievements were made
possible only with large amounts of human supervision. In this work, we aim to
develop unsupervised methods by proposing two methods that create pseudo
query-document pairs and train dense retrieval models in an annotation-free and
scalable manner: query extraction and transferred query generation. The former
method produces pseudo queries by selecting salient spans from the original
document. The latter utilizes generation models trained for other NLP tasks
(e.g., summarization) to produce pseudo queries. Extensive experiments show
that models trained with the proposed augmentation methods can perform
comparably well (or better) to multiple strong baselines. Combining those
strategies leads to further improvements, achieving the state-of-the-art
performance of unsupervised dense retrieval on both BEIR and ODQA datasets. | cs.CL | AugTriever: Unsupervised Dense Retrieval by Scalable Data Augmentation |
2022-12-17T08:34:20Z | http://arxiv.org/pdf/2212.08822v1 | http://arxiv.org/abs/2212.08822v1 | Jiahuan Li, Shanbo Cheng, Zewei Sun, Mingxuan Wang, Shujian Huang | 2022-12-17T08:34:20Z | cs.CL | Nearest Neighbor Machine Translation (kNNMT) is a simple and effective method
of augmenting neural machine translation (NMT) with a token-level nearest
neighbor retrieval mechanism. The effectiveness of kNNMT directly depends on
the quality of retrieved neighbors. However, original kNNMT builds datastores
based on representations from NMT models, which would result in poor retrieval
accuracy when NMT models are not good enough, leading to sub-optimal
translation performance. In this paper, we propose PRED, a framework that
leverages Pre-trained models for Datastores in kNN-MT. Better representations
from pre-trained models allow us to build datastores of better quality. We also
design a novel contrastive alignment objective to mitigate the representation
gap between the NMT model and pre-trained models, enabling the NMT model to
retrieve from better datastores. We conduct extensive experiments on both
bilingual and multilingual translation benchmarks, including WMT17 English
$\leftrightarrow$ Chinese, WMT14 English $\leftrightarrow$ German, IWSLT14
German $\leftrightarrow$ English, and IWSLT14 multilingual datasets. Empirical
results demonstrate the effectiveness of PRED. | cs.CL | Better Datastore, Better Translation: Generating Datastores from Pre-Trained Models for Nearest Neural Machine Translation |
2022-12-17T05:25:17Z | http://arxiv.org/pdf/2212.08802v2 | http://arxiv.org/abs/2212.08802v2 | Bin Wang, Haizhou Li | 2023-06-08T12:44:28Z | cs.CL | We present Relational Sentence Embedding (RSE), a new paradigm to further
discover the potential of sentence embeddings. Prior work mainly models the
similarity between sentences based on their embedding distance. Because of the
complex semantic meanings conveyed, sentence pairs can have various relation
types, including but not limited to entailment, paraphrasing, and
question-answer. It poses challenges to existing embedding methods to capture
such relational information. We handle the problem by learning associated
relational embeddings. Specifically, a relation-wise translation operation is
applied to the source sentence to infer the corresponding target sentence with
a pre-trained Siamese-based encoder. The fine-grained relational similarity
scores can be computed from learned embeddings. We benchmark our method on 19
datasets covering a wide range of tasks, including semantic textual similarity,
transfer, and domain-specific tasks. Experimental results show that our method
is effective and flexible in modeling sentence relations and outperforms a
series of state-of-the-art sentence embedding methods.
https://github.com/BinWang28/RSE | cs.CL | Relational Sentence Embedding for Flexible Semantic Matching |
2022-12-17T02:53:21Z | http://arxiv.org/pdf/2212.08785v1 | http://arxiv.org/abs/2212.08785v1 | Yiyun Zhao, Jiarong Jiang, Yiqun Hu, Wuwei Lan, Henry Zhu, Anuj Chauhan, Alexander Li, Lin Pan, Jun Wang, Chung-Wei Hang, Sheng Zhang, Marvin Dong, Joe Lilien, Patrick Ng, Zhiguo Wang, Vittorio Castelli, Bing Xiang | 2022-12-17T02:53:21Z | cs.CL | Recently, there has been increasing interest in synthesizing data to improve
downstream text-to-SQL tasks. In this paper, we first examined the existing
synthesized datasets and discovered that state-of-the-art text-to-SQL
algorithms did not further improve on popular benchmarks when trained with
augmented synthetic data. We observed two shortcomings: illogical synthetic SQL
queries from independent column sampling and arbitrary table joins. To address
these issues, we propose a novel synthesis framework that incorporates key
relationships from schema, imposes strong typing, and conducts
schema-distance-weighted column sampling. We also adopt an intermediate
representation (IR) for the SQL-to-text task to further improve the quality of
the generated natural language questions. When existing powerful semantic
parsers are pre-finetuned on our high-quality synthesized data, our experiments
show that these models have significant accuracy boosts on popular benchmarks,
including new state-of-the-art performance on Spider. | cs.CL | Importance of Synthesizing High-quality Data for Text-to-SQL Parsing |
2022-12-17T02:20:14Z | http://arxiv.org/pdf/2212.08780v1 | http://arxiv.org/abs/2212.08780v1 | Jifan Chen, Yuhao Zhang, Lan Liu, Rui Dong, Xinchi Chen, Patrick Ng, William Yang Wang, Zhiheng Huang | 2022-12-17T02:20:14Z | cs.CL | There has been great progress in unifying various table-to-text tasks using a
single encoder-decoder model trained via multi-task learning (Xie et al.,
2022). However, existing methods typically encode task information with a
simple dataset name as a prefix to the encoder. This not only limits the
effectiveness of multi-task learning, but also hinders the model's ability to
generalize to new domains or tasks that were not seen during training, which is
crucial for real-world applications. In this paper, we propose compositional
task configurations, a set of prompts prepended to the encoder to improve
cross-task generalization of unified models. We design the task configurations
to explicitly specify the task type, as well as its input and output types. We
show that this not only allows the model to better learn shared knowledge
across different tasks at training, but also allows us to control the model by
composing new configurations that apply novel input-output combinations in a
zero-shot manner. We demonstrate via experiments over ten table-to-text tasks
that our method outperforms the UnifiedSKG baseline by noticeable margins in
both in-domain and zero-shot settings, with average improvements of +0.5 and
+12.6 from using a T5-large backbone, respectively. | cs.CL | Improving Cross-task Generalization of Unified Table-to-text Models with Compositional Task Configurations |
2022-12-17T01:09:22Z | http://arxiv.org/pdf/2212.08775v2 | http://arxiv.org/abs/2212.08775v2 | David Uthus, Jianmo Ni | 2023-05-22T16:53:58Z | cs.CL | Evaluating automatically-generated text summaries is a challenging task.
While there have been many interesting approaches, they still fall short of
human evaluations. We present RISE, a new approach for evaluating summaries by
leveraging techniques from information retrieval. RISE is first trained as a
retrieval task using a dual-encoder retrieval setup, and can then be
subsequently utilized for evaluating a generated summary given an input
document, without gold reference summaries. RISE is especially well suited when
working on new datasets where one may not have reference summaries available
for evaluation. We conduct comprehensive experiments on the SummEval benchmark
(Fabbri et al., 2021) and the results show that RISE has higher correlation
with human evaluations compared to many past approaches to summarization
evaluation. Furthermore, RISE also demonstrates data-efficiency and
generalizability across languages. | cs.CL | RISE: Leveraging Retrieval Techniques for Summarization Evaluation |
2022-12-16T23:37:44Z | http://arxiv.org/pdf/2212.08756v4 | http://arxiv.org/abs/2212.08756v4 | Zhenyuan Lu | 2023-03-16T21:31:25Z | cs.CL, stat.AP | Machine learning models can reach high performance on benchmark natural
language processing (NLP) datasets but fail in more challenging settings. We
study this issue when a pre-trained model learns dataset artifacts in natural
language inference (NLI), the topic of studying the logical relationship
between a pair of text sequences. We provide a variety of techniques for
analyzing and locating dataset artifacts inside the crowdsourced Stanford
Natural Language Inference (SNLI) corpus. We study the stylistic pattern of
dataset artifacts in the SNLI. To mitigate dataset artifacts, we employ a
unique multi-scale data augmentation technique with two distinct frameworks: a
behavioral testing checklist at the sentence level and lexical synonym criteria
at the word level. Specifically, our combination method enhances our model's
resistance to perturbation testing, enabling it to continuously outperform the
pre-trained baseline. | cs.CL | Multi-Scales Data Augmentation Approach In Natural Language Inference For Artifacts Mitigation And Pre-Trained Model Optimization |
2022-12-16T21:44:34Z | http://arxiv.org/pdf/2212.08724v3 | http://arxiv.org/abs/2212.08724v3 | Yuxi Feng, Xiaoyuan Yi, Xiting Wang, Laks V. S. Lakshmanan, Xing Xie | 2023-06-06T04:13:03Z | cs.CL | Self-training (ST) has prospered again in language understanding by
augmenting the fine-tuning of pre-trained language models when labeled data is
insufficient. However, it remains challenging to incorporate ST into
attribute-controllable language generation. Augmented by only self-generated
pseudo text, generation models over-emphasize exploitation of the previously
learned space, suffering from a constrained generalization boundary. We revisit
ST and propose a novel method, DuNST to alleviate this problem. DuNST jointly
models text generation and classification with a shared Variational AutoEncoder
and corrupts the generated pseudo text by two kinds of flexible noise to
disturb the space. In this way, our model could construct and utilize both
pseudo text from given labels and pseudo labels from available unlabeled text,
which are gradually refined during the ST process. We theoretically demonstrate
that DuNST can be regarded as enhancing exploration towards the potential real
text space, providing a guarantee of improved performance. Experiments on three
controllable generation tasks show that DuNST could significantly boost control
accuracy while maintaining comparable generation fluency and diversity against
several strong baselines. | cs.CL | DuNST: Dual Noisy Self Training for Semi-Supervised Controllable Text Generation |
2022-12-16T21:29:41Z | http://arxiv.org/pdf/2212.08718v1 | http://arxiv.org/abs/2212.08718v1 | Anbang Ye, Christopher Cui, Taiwei Shi, Mark O. Riedl | 2022-12-16T21:29:41Z | cs.CL, cs.AI | Automated plot generation is the challenge of generating a sequence of events
that will be perceived by readers as the plot of a coherent story. Traditional
symbolic planners plan a story from a goal state and guarantee logical causal
plot coherence but rely on a library of hand-crafted actions with their
preconditions and effects. This closed world setting limits the length and
diversity of what symbolic planners can generate. On the other hand,
pre-trained neural language models can generate stories with great diversity,
while being generally incapable of ending a story in a specified manner and can
have trouble maintaining coherence. In this paper, we present an approach to
story plot generation that unifies causal planning with neural language models.
We propose to use commonsense knowledge extracted from large language models to
recursively expand a story plot in a backward chaining fashion. Specifically,
our system infers the preconditions for events in the story and then events
that will cause those conditions to become true. We performed automatic
evaluation to measure narrative coherence as indicated by the ability to answer
questions about whether different events in the story are causally related to
other events. Results indicate that our proposed method produces more coherent
plotlines than several strong baselines. | cs.CL | Neural Story Planning |
2022-12-16T20:27:40Z | http://arxiv.org/pdf/2212.08703v1 | http://arxiv.org/abs/2212.08703v1 | Aleksandr Laptev, Boris Ginsburg | 2022-12-16T20:27:40Z | eess.AS, cs.CL, cs.IT, cs.LG, math.IT | This paper presents a class of new fast non-trainable entropy-based
confidence estimation methods for automatic speech recognition. We show how
per-frame entropy values can be normalized and aggregated to obtain a
confidence measure per unit and per word for Connectionist Temporal
Classification (CTC) and Recurrent Neural Network Transducer (RNN-T) models.
Proposed methods have similar computational complexity to the traditional
method based on the maximum per-frame probability, but they are more
adjustable, have a wider effective threshold range, and better push apart the
confidence distributions of correct and incorrect words. We evaluate the
proposed confidence measures on LibriSpeech test sets, and show that they are
up to 2 and 4 times better than confidence estimation based on the maximum
per-frame probability at detecting incorrect words for Conformer-CTC and
Conformer-RNN-T models, respectively. | eess.AS | Fast Entropy-Based Methods of Word-Level Confidence Estimation for End-To-End Automatic Speech Recognition |
2022-12-16T20:01:22Z | http://arxiv.org/pdf/2212.08700v2 | http://arxiv.org/abs/2212.08700v2 | James A. Michaelov, Benjamin K. Bergen | 2023-05-26T07:18:15Z | cs.CL, cs.AI, cs.LG | How well do language models deal with quantification? In this study, we focus
on 'few'-type quantifiers, as in 'few children like toys', which might pose a
particular challenge for language models because the sentence components with
out the quantifier are likely to co-occur, and 'few'-type quantifiers are rare.
We present 960 English sentence stimuli from two human neurolinguistic
experiments to 22 autoregressive transformer models of differing sizes. Not
only do all the models perform poorly on 'few'-type quantifiers, but overall
the larger the model, the worse its performance. This inverse scaling is
consistent with previous work suggesting that larger models increasingly
reflect online rather than offline human processing, and we argue that the
decreasing performance of larger models may challenge uses of language models
as the basis for natural language systems. | cs.CL | Rarely a problem? Language models exhibit inverse scaling in their predictions following few-type quantifiers |
2022-12-16T19:30:01Z | http://arxiv.org/pdf/2212.08686v2 | http://arxiv.org/abs/2212.08686v2 | Yi-Fan Zhang, Hanlin Zhang, Li Erran Li, Eric Xing | 2024-03-28T08:20:12Z | cs.CL | Pre-trained language models (LMs) have shown remarkable reasoning performance
using explanations or chain-of-thoughts (CoT)) for in-context learning. On the
other hand, these reasoning tasks are usually presumed to be more approachable
for symbolic programming. To understand the mechanism of reasoning of LMs, we
curate synthetic datasets containing equivalent (natural, symbolic) data pairs,
where symbolic examples contain first-order logic rules and predicates from
non-parametric knowledge bases (KBs), supporting automated verification of
intermediate reasoning results. Then we revisit neuro-symbolic approaches and
propose to learn from demonstrations containing logic rules and corresponding
examples to iteratively reason over KBs, recovering Prolog's backward chaining
algorithm and supporting automated verification of LMs' outputs. Comprehensive
experiments are included to systematically compare LMLP with CoT in deductive
reasoning settings, showing that LMLP enjoys more than $25\%$ higher accuracy
than CoT on length generalization benchmarks even with smaller model sizes. | cs.CL | Evaluating Step-by-Step Reasoning through Symbolic Verification |
2022-12-16T18:23:43Z | http://arxiv.org/pdf/2212.08635v3 | http://arxiv.org/abs/2212.08635v3 | Junlong Li, Jinyuan Wang, Zhuosheng Zhang, Hai Zhao | 2024-03-28T06:06:59Z | cs.CL, cs.AI | Open-Domain Question Answering (ODQA) aims to answer questions without
explicitly providing specific background documents. This task becomes notably
challenging in a zero-shot setting where no data is available to train tailored
retrieval-reader models. While recent Large Language Models (LLMs) like GPT-3
have demonstrated their effectiveness in zero-shot ODQA using direct prompting
methods, these methods still fall short of fully harnessing the potential of
LLMs when implicitly invoked. In this paper, we propose a Self-Prompting
framework to explicitly utilize the massive knowledge encoded in the parameters
of LLMs and their strong instruction understanding abilities. Concretely, we
prompt LLMs step by step to generate multiple pseudo QA pairs with background
passages and explanations entirely from scratch. These generated elements are
then utilized for in-context learning. Experimental results show that our
method significantly surpasses previous state-of-the-art zero-shot methods on
three widely-used ODQA datasets and even achieves comparable performance with
various customized fine-tuned models on full training data. Our code is
available at https://github.com/lockon-n/self-prompting. | cs.CL | Self-Prompting Large Language Models for Zero-Shot Open-Domain QA |
2022-12-16T18:17:07Z | http://arxiv.org/pdf/2212.09523v1 | http://arxiv.org/abs/2212.09523v1 | Malak Mashaabi, Areej Alotaibi, Hala Qudaih, Raghad Alnashwan, Hend Al-Khalifa | 2022-12-16T18:17:07Z | cs.CL, cs.AI | Artificial intelligence and natural language processing (NLP) are
increasingly being used in customer service to interact with users and answer
their questions. The goal of this systematic review is to examine existing
research on the use of NLP technology in customer service, including the
research domain, applications, datasets used, and evaluation methods. The
review also looks at the future direction of the field and any significant
limitations. The review covers the time period from 2015 to 2022 and includes
papers from five major scientific databases. Chatbots and question-answering
systems were found to be used in 10 main fields, with the most common use in
general, social networking, and e-commerce areas. Twitter was the second most
commonly used dataset, with most research also using their own original
datasets. Accuracy, precision, recall, and F1 were the most common evaluation
methods. Future work aims to improve the performance and understanding of user
behavior and emotions, and address limitations such as the volume, diversity,
and quality of datasets. This review includes research on different spoken
languages and models and techniques. | cs.CL | Natural Language Processing in Customer Service: A Systematic Review |
2022-12-16T18:12:04Z | http://arxiv.org/pdf/2212.08632v2 | http://arxiv.org/abs/2212.08632v2 | Qian Yang, Qian Chen, Wen Wang, Baotian Hu, Min Zhang | 2023-08-07T03:02:06Z | cs.CL, cs.CV | Multi-modal multi-hop question answering involves answering a question by
reasoning over multiple input sources from different modalities. Existing
methods often retrieve evidences separately and then use a language model to
generate an answer based on the retrieved evidences, and thus do not adequately
connect candidates and are unable to model the interdependent relations during
retrieval. Moreover, the pipelined approaches of retrieval and generation might
result in poor generation performance when retrieval performance is low. To
address these issues, we propose a Structured Knowledge and Unified
Retrieval-Generation (SKURG) approach. SKURG employs an Entity-centered Fusion
Encoder to align sources from different modalities using shared entities. It
then uses a unified Retrieval-Generation Decoder to integrate intermediate
retrieval results for answer generation and also adaptively determine the
number of retrieval steps. Extensive experiments on two representative
multi-modal multi-hop QA datasets MultimodalQA and WebQA demonstrate that SKURG
outperforms the state-of-the-art models in both source retrieval and answer
generation performance with fewer parameters. Our code is available at
https://github.com/HITsz-TMG/SKURG. | cs.CL | Enhancing Multi-modal and Multi-hop Question Answering via Structured Knowledge and Unified Retrieval-Generation |
2022-12-16T17:57:41Z | http://arxiv.org/pdf/2212.08620v2 | http://arxiv.org/abs/2212.08620v2 | Jiaxin Pei, Aparna Ananthasubramaniam, Xingyao Wang, Naitian Zhou, Jackson Sargent, Apostolos Dedeloudis, David Jurgens | 2023-03-23T18:45:37Z | cs.CL, cs.AI, cs.CY, cs.HC, cs.LG | We present POTATO, the Portable text annotation tool, a free, fully
open-sourced annotation system that 1) supports labeling many types of text and
multimodal data; 2) offers easy-to-configure features to maximize the
productivity of both deployers and annotators (convenient templates for common
ML/NLP tasks, active learning, keypress shortcuts, keyword highlights,
tooltips); and 3) supports a high degree of customization (editable UI,
inserting pre-screening questions, attention and qualification tests).
Experiments over two annotation tasks suggest that POTATO improves labeling
speed through its specially-designed productivity features, especially for long
documents and complex tasks. POTATO is available at
https://github.com/davidjurgens/potato and will continue to be updated. | cs.CL | POTATO: The Portable Text Annotation Tool |
2022-12-16T17:57:14Z | http://arxiv.org/pdf/2212.08619v1 | http://arxiv.org/abs/2212.08619v1 | C. M. Downey, Wei Dai, Huseyin A. Inan, Kim Laine, Saurabh Naik, Tomasz Religa | 2022-12-16T17:57:14Z | cs.CL, cs.CR | Language models are widely deployed to provide automatic text completion
services in user products. However, recent research has revealed that language
models (especially large ones) bear considerable risk of memorizing private
training data, which is then vulnerable to leakage and extraction by
adversaries. In this study, we test the efficacy of a range of
privacy-preserving techniques to mitigate unintended memorization of sensitive
user text, while varying other factors such as model size and adversarial
conditions. We test both "heuristic" mitigations (those without formal privacy
guarantees) and Differentially Private training, which provides provable levels
of privacy at the cost of some model performance. Our experiments show that
(with the exception of L2 regularization), heuristic mitigations are largely
ineffective in preventing memorization in our test suite, possibly because they
make too strong of assumptions about the characteristics that define
"sensitive" or "private" text. In contrast, Differential Privacy reliably
prevents memorization in our experiments, despite its computational and
model-performance costs. | cs.CL | Planting and Mitigating Memorized Content in Predictive-Text Language Models |
2022-12-16T17:36:23Z | http://arxiv.org/pdf/2212.08607v1 | http://arxiv.org/abs/2212.08607v1 | Swarnadeep Saha, Xinyan Velocity Yu, Mohit Bansal, Ramakanth Pasunuru, Asli Celikyilmaz | 2022-12-16T17:36:23Z | cs.CL, cs.AI, cs.LG | Prompting large language models has enabled significant recent progress in
multi-step reasoning over text. However, when applied to text generation from
semi-structured data (e.g., graphs or tables), these methods typically suffer
from low semantic coverage, hallucination, and logical inconsistency. We
propose MURMUR, a neuro-symbolic modular approach to text generation from
semi-structured data with multi-step reasoning. MURMUR is a best-first search
method that generates reasoning paths using: (1) neural and symbolic modules
with specific linguistic and logical skills, (2) a grammar whose production
rules define valid compositions of modules, and (3) value functions that assess
the quality of each reasoning step. We conduct experiments on two diverse
data-to-text generation tasks like WebNLG and LogicNLG. These tasks differ in
their data representations (graphs and tables) and span multiple linguistic and
logical skills. MURMUR obtains significant improvements over recent few-shot
baselines like direct prompting and chain-of-thought prompting, while also
achieving comparable performance to fine-tuned GPT-2 on out-of-domain data.
Moreover, human evaluation shows that MURMUR generates highly faithful and
correct reasoning paths that lead to 26% more logically consistent summaries on
LogicNLG, compared to direct prompting. | cs.CL | MURMUR: Modular Multi-Step Reasoning for Semi-Structured Data-to-Text Generation |
2022-12-16T17:24:49Z | http://arxiv.org/pdf/2212.08597v2 | http://arxiv.org/abs/2212.08597v2 | David Dale, Elena Voita, Loïc Barrault, Marta R. Costa-jussà | 2022-12-20T16:11:02Z | cs.CL, I.2.7 | While the problem of hallucinations in neural machine translation has long
been recognized, so far the progress on its alleviation is very little. Indeed,
recently it turned out that without artificially encouraging models to
hallucinate, previously existing methods fall short and even the standard
sequence log-probability is more informative. It means that characteristics
internal to the model can give much more information than we expect, and before
using external models and measures, we first need to ask: how far can we go if
we use nothing but the translation model itself ? We propose to use a method
that evaluates the percentage of the source contribution to a generated
translation. Intuitively, hallucinations are translations "detached" from the
source, hence they can be identified by low source contribution. This method
improves detection accuracy for the most severe hallucinations by a factor of 2
and is able to alleviate hallucinations at test time on par with the previous
best approach that relies on external models. Next, if we move away from
internal model characteristics and allow external tools, we show that using
sentence similarity from cross-lingual embeddings further improves these
results. | cs.CL | Detecting and Mitigating Hallucinations in Machine Translation: Model Internal Workings Alone Do Well, Sentence Similarity Even Better |
2022-12-16T16:00:19Z | http://arxiv.org/pdf/2212.08550v1 | http://arxiv.org/abs/2212.08550v1 | Matyáš Boháček, Michal Bravanský, Filip Trhlík, Václav Moravec | 2022-12-16T16:00:19Z | cs.CL, I.2.7; K.4.0 | We present the Verifee Dataset: a novel dataset of news articles with
fine-grained trustworthiness annotations. We develop a detailed methodology
that assesses the texts based on their parameters encompassing editorial
transparency, journalist conventions, and objective reporting while penalizing
manipulative techniques. We bring aboard a diverse set of researchers from
social, media, and computer sciences to overcome barriers and limited framing
of this interdisciplinary problem. We collect over $10,000$ unique articles
from almost $60$ Czech online news sources. These are categorized into one of
the $4$ classes across the credibility spectrum we propose, raging from
entirely trustworthy articles all the way to the manipulative ones. We produce
detailed statistics and study trends emerging throughout the set. Lastly, we
fine-tune multiple popular sequence-to-sequence language models using our
dataset on the trustworthiness classification task and report the best testing
F-1 score of $0.52$. We open-source the dataset, annotation methodology, and
annotators' instructions in full length at https://verifee.ai/research to
enable easy build-up work. We believe similar methods can help prevent
disinformation and educate in the realm of media literacy. | cs.CL | Fine-grained Czech News Article Dataset: An Interdisciplinary Approach to Trustworthiness Analysis |
2022-12-16T15:46:15Z | http://arxiv.org/pdf/2212.08542v2 | http://arxiv.org/abs/2212.08542v2 | Suwon Shon, Felix Wu, Kwangyoun Kim, Prashant Sridhar, Karen Livescu, Shinji Watanabe | 2023-03-28T21:20:11Z | eess.AS, cs.CL | Self-supervised pre-trained transformers have improved the state of the art
on a variety of speech tasks. Due to the quadratic time and space complexity of
self-attention, they usually operate at the level of relatively short (e.g.,
utterance) segments. In this paper, we study the use of context, i.e.,
surrounding segments, during fine-tuning and propose a new approach called
context-aware fine-tuning. We attach a context module on top of the last layer
of a pre-trained model to encode the whole segment into a context embedding
vector which is then used as an additional feature for the final prediction.
During the fine-tuning stage, we introduce an auxiliary loss that encourages
this context embedding vector to be similar to context vectors of surrounding
segments. This allows the model to make predictions without access to these
surrounding segments at inference time and requires only a tiny overhead
compared to standard fine-tuned models. We evaluate the proposed approach using
the SLUE and Libri-light benchmarks for several downstream tasks: Automatic
speech recognition (ASR), named entity recognition (NER), and sentiment
analysis (SA). The results show that context-aware fine-tuning not only
outperforms a standard fine-tuning baseline but also rivals a strong context
injection baseline that uses neighboring speech segments during inference. | eess.AS | Context-aware Fine-tuning of Self-supervised Speech Models |
2022-12-16T14:54:56Z | http://arxiv.org/pdf/2212.08514v1 | http://arxiv.org/abs/2212.08514v1 | Amani S. Abumansour, Arkaitz Zubiaga | 2022-12-16T14:54:56Z | cs.CL | An important component of an automated fact-checking system is the claim
check-worthiness detection system, which ranks sentences by prioritising them
based on their need to be checked. Despite a body of research tackling the
task, previous research has overlooked the challenging nature of identifying
check-worthy claims across different topics. In this paper, we assess and
quantify the challenge of detecting check-worthy claims for new, unseen topics.
After highlighting the problem, we propose the AraCWA model to mitigate the
performance deterioration when detecting check-worthy claims across topics. The
AraCWA model enables boosting the performance for new topics by incorporating
two components for few-shot learning and data augmentation. Using a publicly
available dataset of Arabic tweets consisting of 14 different topics, we
demonstrate that our proposed data augmentation strategy achieves substantial
improvements across topics overall, where the extent of the improvement varies
across topics. Further, we analyse the semantic similarities between topics,
suggesting that the similarity metric could be used as a proxy to determine the
difficulty level of an unseen topic prior to undertaking the task of labelling
the underlying sentences. | cs.CL | Check-worthy Claim Detection across Topics for Automated Fact-checking |
2022-12-16T14:01:42Z | http://arxiv.org/pdf/2212.08489v2 | http://arxiv.org/abs/2212.08489v2 | Esaú Villatoro-Tello, Srikanth Madikeri, Juan Zuluaga-Gomez, Bidisha Sharma, Seyyed Saeed Sarfjoo, Iuliia Nigmatulina, Petr Motlicek, Alexei V. Ivanov, Aravind Ganapathiraju | 2023-03-17T13:26:14Z | cs.CL, cs.AI, cs.SD, eess.AS, I.2.7 | In this paper, we perform an exhaustive evaluation of different
representations to address the intent classification problem in a Spoken
Language Understanding (SLU) setup. We benchmark three types of systems to
perform the SLU intent detection task: 1) text-based, 2) lattice-based, and a
novel 3) multimodal approach. Our work provides a comprehensive analysis of
what could be the achievable performance of different state-of-the-art SLU
systems under different circumstances, e.g., automatically- vs.
manually-generated transcripts. We evaluate the systems on the publicly
available SLURP spoken language resource corpus. Our results indicate that
using richer forms of Automatic Speech Recognition (ASR) outputs, namely
word-consensus-networks, allows the SLU system to improve in comparison to the
1-best setup (5.5% relative improvement). However, crossmodal approaches, i.e.,
learning from acoustic and text embeddings, obtains performance similar to the
oracle setup, a relative improvement of 17.8% over the 1-best configuration,
being a recommended alternative to overcome the limitations of working with
automatically generated transcripts. | cs.CL | Effectiveness of Text, Acoustic, and Lattice-based representations in Spoken Language Understanding tasks |
2022-12-16T14:00:26Z | http://arxiv.org/pdf/2212.08486v1 | http://arxiv.org/abs/2212.08486v1 | Mingda Chen, Paul-Ambroise Duquenne, Pierre Andrews, Justine Kao, Alexandre Mourachko, Holger Schwenk, Marta R. Costa-jussà | 2022-12-16T14:00:26Z | cs.CL, I.2.7 | End-to-End speech-to-speech translation (S2ST) is generally evaluated with
text-based metrics. This means that generated speech has to be automatically
transcribed, making the evaluation dependent on the availability and quality of
automatic speech recognition (ASR) systems. In this paper, we propose a
text-free evaluation metric for end-to-end S2ST, named BLASER, to avoid the
dependency on ASR systems. BLASER leverages a multilingual multimodal encoder
to directly encode the speech segments for source input, translation output and
reference into a shared embedding space and computes a score of the translation
quality that can be used as a proxy to human evaluation. To evaluate our
approach, we construct training and evaluation sets from more than 40k human
annotations covering seven language directions. The best results of BLASER are
achieved by training with supervision from human rating scores. We show that
when evaluated at the sentence level, BLASER correlates significantly better
with human judgment compared to ASR-dependent metrics including ASR-SENTBLEU in
all translation directions and ASR-COMET in five of them. Our analysis shows
combining speech and text as inputs to BLASER does not increase the correlation
with human scores, but best correlations are achieved when using speech, which
motivates the goal of our research. Moreover, we show that using ASR for
references is detrimental for text-based metrics. | cs.CL | BLASER: A Text-Free Speech-to-Speech Translation Evaluation Metric |
2022-12-16T13:55:22Z | http://arxiv.org/pdf/2212.08482v2 | http://arxiv.org/abs/2212.08482v2 | Iosif Iulian Petrila | 2022-12-22T09:04:07Z | cs.CL, cs.FL, cs.PL, cs.SC | The general translator formalism and computing specific implementations are
proposed. The implementation of specific elements necessary to process the
source and destination information within the translators are presented. Some
common directives or instructions, such as classes and procedures, were unified
and generalized in order to allow general translations implementations. In
order to cover general cases, two levels of processing are required, related to
the source and destination information appropriate transformations, with the
related control and processing instructions. The proposed general translator
elements are useful for processing natural or artificial information described
through any types of languages or systems. | cs.CL | Implementation of general formal translators |
2022-12-16T13:07:39Z | http://arxiv.org/pdf/2212.08459v1 | http://arxiv.org/abs/2212.08459v1 | Muriël de Groot, Mohammad Aliannejadi, Marcel R. Haas | 2022-12-16T13:07:39Z | cs.CL | Topic modeling is widely used for analytically evaluating large collections
of textual data. One of the most popular topic techniques is Latent Dirichlet
Allocation (LDA), which is flexible and adaptive, but not optimal for e.g.
short texts from various domains. We explore how the state-of-the-art BERTopic
algorithm performs on short multi-domain text and find that it generalizes
better than LDA in terms of topic coherence and diversity. We further analyze
the performance of the HDBSCAN clustering algorithm utilized by BERTopic and
find that it classifies a majority of the documents as outliers. This crucial,
yet overseen problem excludes too many documents from further analysis. When we
replace HDBSCAN with k-Means, we achieve similar performance, but without
outliers. | cs.CL | Experiments on Generalizability of BERTopic on Multi-Domain Short Text |
2022-12-16T13:07:09Z | http://arxiv.org/pdf/2212.08458v1 | http://arxiv.org/abs/2212.08458v1 | Tianyu Shi, Zhicheng Wang, Liyin Xiao, Cong Liu | 2022-12-16T13:07:09Z | cs.CL, cs.AI, cs.LG | Most recent studies on neural constituency parsing focus on encoder
structures, while few developments are devoted to decoders. Previous research
has demonstrated that probabilistic statistical methods based on syntactic
rules are particularly effective in constituency parsing, whereas syntactic
rules are not used during the training of neural models in prior work probably
due to their enormous computation requirements. In this paper, we first
implement a fast CKY decoding procedure harnessing GPU acceleration, based on
which we further derive a syntactic rule-based (rule-constrained) CKY decoding.
In the experiments, our method obtains 95.89 and 92.52 F1 on the datasets of
PTB and CTB respectively, which shows significant improvements compared with
previous approaches. Besides, our parser achieves strong and competitive
cross-domain performance in zero-shot settings. | cs.CL | Fast Rule-Based Decoding: Revisiting Syntactic Rules in Neural Constituency Parsing |
2022-12-16T11:34:49Z | http://arxiv.org/pdf/2212.14672v1 | http://arxiv.org/abs/2212.14672v1 | Yuyang Chen, Xiaoyu Cui, Yunjie Song, Manli Wu | 2022-12-16T11:34:49Z | cs.CY, cs.CL, cs.SI | This study verified the effectiveness of Donald Trump's Twitter campaign in
guiding agen-da-setting and deflecting political risk and examined Trump's
Twitter communication strategy and explores the communication effects of his
tweet content during Covid-19 pandemic. We collected all tweets posted by Trump
on the Twitter platform from January 1, 2020 to December 31, 2020.We used
Ordinary Least Squares (OLS) regression analysis with a fixed effects model to
analyze the existence of the Twitter strategy. The correlation between the
number of con-firmed daily Covid-19 diagnoses and the number of particular
thematic tweets was investigated using time series analysis. Empirical analysis
revealed Twitter's strategy is used to divert public attention from negative
Covid-19 reports during the epidemic, and it posts a powerful political
communication effect on Twitter. However, findings suggest that Trump did not
use false claims to divert political risk and shape public opinion. | cs.CY | Twitter's Agenda-Setting Role: A Study of Twitter Strategy for Political Diversion |
2022-12-16T11:24:42Z | http://arxiv.org/pdf/2212.08410v3 | http://arxiv.org/abs/2212.08410v3 | Lucie Charlotte Magister, Jonathan Mallinson, Jakub Adamek, Eric Malmi, Aliaksei Severyn | 2023-06-01T12:17:01Z | cs.CL, cs.LG | Chain of thought prompting successfully improves the reasoning capabilities
of large language models, achieving state of the art results on a range of
datasets. However, these reasoning capabilities only appear to emerge in models
with a size of over 100 billion parameters. In this paper, we explore the
transfer of such reasoning capabilities to models with less than 100 billion
parameters via knowledge distillation. Specifically, we finetune a student
model on the chain of thought outputs generated by a larger teacher model. Our
experiments show that the proposed method improves task performance across
arithmetic, commonsense and symbolic reasoning datasets. For example, the
accuracy of T5 XXL on GSM8K improves from 8.11% to 21.99% when finetuned on
PaLM-540B generated chains of thought. | cs.CL | Teaching Small Language Models to Reason |
2022-12-16T11:15:39Z | http://arxiv.org/pdf/2212.08408v2 | http://arxiv.org/abs/2212.08408v2 | Ganqu Cui, Wentao Li, Ning Ding, Longtao Huang, Zhiyuan Liu, Maosong Sun | 2023-05-24T14:30:50Z | cs.CL, cs.LG | With the evergrowing sizes of pre-trained models (PTMs), it has been an
emerging practice to only provide the inference APIs for users, namely
model-as-a-service (MaaS) setting. To adapt PTMs with model parameters frozen,
most current approaches focus on the input side, seeking for powerful prompts
to stimulate models for correct answers. However, we argue that input-side
adaptation could be arduous due to the lack of gradient signals and they
usually require thousands of API queries, resulting in high computation and
time costs. In light of this, we present Decoder Tuning (DecT), which in
contrast optimizes task-specific decoder networks on the output side.
Specifically, DecT first extracts prompt-stimulated output scores for initial
predictions. On top of that, we train an additional decoder network on the
output representations to incorporate posterior data knowledge. By
gradient-based optimization, DecT can be trained within several seconds and
requires only one PTM query per sample. Empirically, we conduct extensive
natural language understanding experiments and show that DecT significantly
outperforms state-of-the-art algorithms with a $200\times$ speed-up. | cs.CL | Decoder Tuning: Efficient Language Understanding as Decoding |
2022-12-16T11:06:48Z | http://arxiv.org/pdf/2212.08407v1 | http://arxiv.org/abs/2212.08407v1 | Fatemeh Sadat Masoumi, Mohammad Bahrani | 2022-12-16T11:06:48Z | cs.CL, cs.CY | The COVID-19 pandemic has caused drastic alternations in human life in all
aspects. The government's laws in this regard affected the lifestyle of all
people. Due to this fact studying the sentiment of individuals is essential to
be aware of the future impacts of the coming pandemics. To contribute to this
aim, we proposed an NLP (Natural Language Processing) model to analyze
open-text answers in a survey in Persian and detect positive and negative
feelings of the people in Iran. In this study, a distilBert transformer model
was applied to take on this task. We deployed three approaches to perform the
comparison, and our best model could gain accuracy: 0.824, Precision: 0.824,
Recall: 0.798, and F1 score: 0.804. | cs.CL | Utilizing distilBert transformer model for sentiment classification of COVID-19's Persian open-text responses |
2022-12-16T10:39:22Z | http://arxiv.org/pdf/2212.08395v1 | http://arxiv.org/abs/2212.08395v1 | Rowan Hall Maudslay, Simone Teufel | 2022-12-16T10:39:22Z | cs.CL | Linguists distinguish between novel and conventional metaphor, a distinction
which the metaphor detection task in NLP does not take into account. Instead,
metaphoricity is formulated as a property of a token in a sentence, regardless
of metaphor type. In this paper, we investigate the limitations of treating
conventional metaphors in this way, and advocate for an alternative which we
name 'metaphorical polysemy detection' (MPD). In MPD, only conventional
metaphoricity is treated, and it is formulated as a property of word senses in
a lexicon. We develop the first MPD model, which learns to identify
conventional metaphors in the English WordNet. To train it, we present a novel
training procedure that combines metaphor detection with word sense
disambiguation (WSD). For evaluation, we manually annotate metaphor in two
subsets of WordNet. Our model significantly outperforms a strong baseline based
on a state-of-the-art metaphor detection model, attaining an ROC-AUC score of
.78 (compared to .65) on one of the sets. Additionally, when paired with a WSD
model, our approach outperforms a state-of-the-art metaphor detection model at
identifying conventional metaphors in text (.659 F1 compared to .626). | cs.CL | Metaphorical Polysemy Detection: Conventional Metaphor meets Word Sense Disambiguation |
2022-12-16T10:33:38Z | http://arxiv.org/pdf/2212.08390v2 | http://arxiv.org/abs/2212.08390v2 | Rodrigo Agerri, Eneko Agirre | 2023-09-22T07:55:52Z | cs.CL, cs.AI | Given the impact of language models on the field of Natural Language
Processing, a number of Spanish encoder-only masked language models (aka BERTs)
have been trained and released. These models were developed either within large
projects using very large private corpora or by means of smaller scale academic
efforts leveraging freely available data. In this paper we present a
comprehensive head-to-head comparison of language models for Spanish with the
following results: (i) Previously ignored multilingual models from large
companies fare better than monolingual models, substantially changing the
evaluation landscape of language models in Spanish; (ii) Results across the
monolingual models are not conclusive, with supposedly smaller and inferior
models performing competitively. Based on these empirical results, we argue for
the need of more research to understand the factors underlying them. In this
sense, the effect of corpus size, quality and pre-training techniques need to
be further investigated to be able to obtain Spanish monolingual models
significantly better than the multilingual ones released by large private
companies, specially in the face of rapid ongoing progress in the field. The
recent activity in the development of language technology for Spanish is to be
welcomed, but our results show that building language models remains an open,
resource-heavy problem which requires to marry resources (monetary and/or
computational) with the best research expertise and practice. | cs.CL | Lessons learned from the evaluation of Spanish Language Models |
2022-12-16T10:23:26Z | http://arxiv.org/pdf/2212.08388v1 | http://arxiv.org/abs/2212.08388v1 | Rowan Hall Maudslay, Simone Teufel | 2022-12-16T10:23:26Z | cs.CL | A widely acknowledged shortcoming of WordNet is that it lacks a distinction
between word meanings which are systematically related (polysemy), and those
which are coincidental (homonymy). Several previous works have attempted to
fill this gap, by inferring this information using computational methods. We
revisit this task, and exploit recent advances in language modelling to
synthesise homonymy annotation for Princeton WordNet. Previous approaches treat
the problem using clustering methods; by contrast, our method works by linking
WordNet to the Oxford English Dictionary, which contains the information we
need. To perform this alignment, we pair definitions based on their proximity
in an embedding space produced by a Transformer model. Despite the simplicity
of this approach, our best model attains an F1 of .97 on an evaluation set that
we annotate. The outcome of our work is a high-quality homonymy annotation
layer for Princeton WordNet, which we release. | cs.CL | Homonymy Information for English WordNet |
2022-12-16T09:01:56Z | http://arxiv.org/pdf/2212.08354v1 | http://arxiv.org/abs/2212.08354v1 | Weilong Dong, Xinwei Wu, Junzhuo Li, Shuangzhi Wu, Chao Bian, Deyi Xiong | 2022-12-16T09:01:56Z | cs.CL | Massively multi-task learning with large language models has recently made
substantial progress on few-shot generalization. However, this is usually
performed in a centralized learning fashion, ignoring the privacy sensitivity
issue of (annotated) data used in multiple tasks. To mitigate this issue, we
propose FewFedWeight, a few-shot federated learning framework across multiple
tasks, to achieve the best of both worlds: privacy preservation and cross-task
generalization. FewFedWeight trains client models in isolated devices without
sharing data. It broadcasts the global model in the server to each client and
produces pseudo data for clients so that knowledge from the global model can be
explored to enhance few-shot learning of each client model. An energy-based
algorithm is further proposed to weight pseudo samples in order to reduce the
negative impact of noise from the generated pseudo data. Adaptive model weights
of client models are also tuned according to their performance. We use these
model weights to dynamically aggregate client models to update the global
model. Experiments on 118 NLP tasks show that FewFedWeight can significantly
improve the performance of client models on 61% tasks with an average
performance improvement rate of 30.5% over the baseline and substantially
outperform FedAvg and other decentralized learning methods. | cs.CL | FewFedWeight: Few-shot Federated Learning Framework across Multiple NLP Tasks |
2022-12-16T09:01:19Z | http://arxiv.org/pdf/2212.08353v1 | http://arxiv.org/abs/2212.08353v1 | Christine de Kock, Tom Stafford, Andreas Vlachos | 2022-12-16T09:01:19Z | cs.CL, cs.AI, cs.CY | Disagreements are frequently studied from the perspective of either detecting
toxicity or analysing argument structure. We propose a framework of dispute
tactics that unifies these two perspectives, as well as other dialogue acts
which play a role in resolving disputes, such as asking questions and providing
clarification. This framework includes a preferential ordering among
rebuttal-type tactics, ranging from ad hominem attacks to refuting the central
argument. Using this framework, we annotate 213 disagreements (3,865
utterances) from Wikipedia Talk pages. This allows us to investigate research
questions around the tactics used in disagreements; for instance, we provide
empirical validation of the approach to disagreement recommended by Wikipedia.
We develop models for multilabel prediction of dispute tactics in an utterance,
achieving the best performance with a transformer-based label powerset model.
Adding an auxiliary task to incorporate the ordering of rebuttal tactics
further yields a statistically significant increase. Finally, we show that
these annotations can be used to provide useful additional signals to improve
performance on the task of predicting escalation. | cs.CL | How to disagree well: Investigating the dispute tactics used on Wikipedia |
2022-12-16T08:57:18Z | http://arxiv.org/pdf/2212.08349v1 | http://arxiv.org/abs/2212.08349v1 | Junzhuo Li, Xinwei Wu, Weilong Dong, Shuangzhi Wu, Chao Bian, Deyi Xiong | 2022-12-16T08:57:18Z | cs.LG, cs.AI, cs.CL, cs.CR | Knowledge distillation (KD) has been widely used for model compression and
knowledge transfer. Typically, a big teacher model trained on sufficient data
transfers knowledge to a small student model. However, despite the success of
KD, little effort has been made to study whether KD leaks the training data of
the teacher model. In this paper, we experimentally reveal that KD suffers from
the risk of privacy leakage. To alleviate this issue, we propose a novel
knowledge distillation method, swing distillation, which can effectively
protect the private information of the teacher model from flowing to the
student model. In our framework, the temperature coefficient is dynamically and
adaptively adjusted according to the degree of private information contained in
the data, rather than a predefined constant hyperparameter. It assigns
different temperatures to tokens according to the likelihood that a token in a
position contains private information. In addition, we inject noise into soft
targets provided to the student model, in order to avoid unshielded knowledge
transfer. Experiments on multiple datasets and tasks demonstrate that the
proposed swing distillation can significantly reduce (by over 80% in terms of
canary exposure) the risk of privacy leakage in comparison to KD with
competitive or better performance. Furthermore, swing distillation is robust
against the increasing privacy budget. | cs.LG | Swing Distillation: A Privacy-Preserving Knowledge Distillation Framework |
2022-12-16T08:26:32Z | http://arxiv.org/pdf/2212.08335v1 | http://arxiv.org/abs/2212.08335v1 | Ha-Thanh Nguyen, Vu Tran, Ngoc-Cam Le, Thi-Thuy Le, Quang-Huy Nguyen, Le-Minh Nguyen, Ken Satoh | 2022-12-16T08:26:32Z | cs.CL | Knowledge representation and reasoning in law are essential to facilitate the
automation of legal analysis and decision-making tasks. In this paper, we
propose a new approach based on legal science, specifically legal taxonomy, for
representing and reasoning with legal documents. Our approach interprets the
regulations in legal documents as binary trees, which facilitates legal
reasoning systems to make decisions and resolve logical contradictions. The
advantages of this approach are twofold. First, legal reasoning can be
performed on the basis of the binary tree representation of the regulations.
Second, the binary tree representation of the regulations is more
understandable than the existing sentence-based representations. We provide an
example of how our approach can be used to interpret the regulations in a legal
document. | cs.CL | Law to Binary Tree -- An Formal Interpretation of Legal Natural Language |
2022-12-16T08:14:04Z | http://arxiv.org/pdf/2212.08329v1 | http://arxiv.org/abs/2212.08329v1 | Yusuke Yasuda, Tomoki Toda | 2022-12-16T08:14:04Z | eess.AS, cs.CL, stat.ML | Text-to-speech synthesis (TTS) is a task to convert texts into speech. Two of
the factors that have been driving TTS are the advancements of probabilistic
models and latent representation learning. We propose a TTS method based on
latent variable conversion using a diffusion probabilistic model and the
variational autoencoder (VAE). In our TTS method, we use a waveform model based
on VAE, a diffusion model that predicts the distribution of latent variables in
the waveform model from texts, and an alignment model that learns alignments
between the text and speech latent sequences. Our method integrates diffusion
with VAE by modeling both mean and variance parameters with diffusion, where
the target distribution is determined by approximation from VAE. This latent
variable conversion framework potentially enables us to flexibly incorporate
various latent feature extractors. Our experiments show that our method is
robust to linguistic labels with poor orthography and alignment errors. | eess.AS | Text-to-speech synthesis based on latent variable conversion using diffusion probabilistic model and variational autoencoder |
2022-12-16T08:14:04Z | http://arxiv.org/pdf/2212.08330v2 | http://arxiv.org/abs/2212.08330v2 | Yujing Wang, Yaming Yang, Zhuo Li, Jiangang Bai, Mingliang Zhang, Xiangtai Li, Jing Yu, Ce Zhang, Gao Huang, Yunhai Tong | 2023-04-28T06:44:48Z | cs.LG, cs.CL, cs.CV, cs.NE | Attention-based neural networks, such as Transformers, have become ubiquitous
in numerous applications, including computer vision, natural language
processing, and time-series analysis. In all kinds of attention networks, the
attention maps are crucial as they encode semantic dependencies between input
tokens. However, most existing attention networks perform modeling or reasoning
based on representations , wherein the attention maps of different layers are
learned separately without explicit interactions. In this paper, we propose a
novel and generic evolving attention mechanism, which directly models the
evolution of inter-token relationships through a chain of residual
convolutional modules. The major motivations are twofold. On the one hand, the
attention maps in different layers share transferable knowledge, thus adding a
residual connection can facilitate the information flow of inter-token
relationships across layers. On the other hand, there is naturally an
evolutionary trend among attention maps at different abstraction levels, so it
is beneficial to exploit a dedicated convolution-based module to capture this
process. Equipped with the proposed mechanism, the convolution-enhanced
evolving attention networks achieve superior performance in various
applications, including time-series representation, natural language
understanding, machine translation, and image classification. Especially on
time-series representation tasks, Evolving Attention-enhanced Dilated
Convolutional (EA-DC-) Transformer outperforms state-of-the-art models
significantly, achieving an average of 17% improvement compared to the best
SOTA. To the best of our knowledge, this is the first work that explicitly
models the layer-wise evolution of attention maps. Our implementation is
available at https://github.com/pkuyym/EvolvingAttention. | cs.LG | Convolution-enhanced Evolving Attention Networks |
2022-12-16T08:07:51Z | http://arxiv.org/pdf/2212.11219v2 | http://arxiv.org/abs/2212.11219v2 | Bharath Muppasani, Vishal Pallagani, Kausik Lakkaraju, Shuge Lei, Biplav Srivastava, Brett Robertson, Andrea Hickerson, Vignesh Narayanan | 2022-12-28T08:09:45Z | cs.HC, cs.CL, cs.CY | Chatbots, or bots for short, are multi-modal collaborative assistants that
can help people complete useful tasks. Usually, when chatbots are referenced in
connection with elections, they often draw negative reactions due to the fear
of mis-information and hacking. Instead, in this paper, we explore how chatbots
may be used to promote voter participation in vulnerable segments of society
like senior citizens and first-time voters. In particular, we build a system
that amplifies official information while personalizing it to users' unique
needs transparently. We discuss its design, build prototypes with frequently
asked questions (FAQ) election information for two US states that are low on an
ease-of-voting scale, and report on its initial evaluation in a focus group.
Our approach can be a win-win for voters, election agencies trying to fulfill
their mandate and democracy at large. | cs.HC | On Safe and Usable Chatbots for Promoting Voter Participation |
2022-12-16T07:48:02Z | http://arxiv.org/pdf/2212.08322v1 | http://arxiv.org/abs/2212.08322v1 | Kai Xiong, Xiao Ding, Zhongyang Li, Li Du, Bing Qin, Yi Zheng, Baoxing Huai | 2022-12-16T07:48:02Z | cs.AI, cs.CL | Causal chain reasoning (CCR) is an essential ability for many decision-making
AI systems, which requires the model to build reliable causal chains by
connecting causal pairs. However, CCR suffers from two main transitive
problems: threshold effect and scene drift. In other words, the causal pairs to
be spliced may have a conflicting threshold boundary or scenario. To address
these issues, we propose a novel Reliable Causal chain reasoning
framework~(ReCo), which introduces exogenous variables to represent the
threshold and scene factors of each causal pair within the causal chain, and
estimates the threshold and scene contradictions across exogenous variables via
structural causal recurrent neural networks~(SRNN). Experiments show that ReCo
outperforms a series of strong baselines on both Chinese and English CCR
datasets. Moreover, by injecting reliable causal chain knowledge distilled by
ReCo, BERT can achieve better performances on four downstream causal-related
tasks than BERT models enhanced by other kinds of knowledge. | cs.AI | ReCo: Reliable Causal Chain Reasoning via Structural Causal Recurrent Neural Networks |
2022-12-16T07:47:03Z | http://arxiv.org/pdf/2212.08321v1 | http://arxiv.org/abs/2212.08321v1 | Yusuke Yasuda, Tomoki Toda | 2022-12-16T07:47:03Z | eess.AS, cs.CL | End-to-end text-to-speech synthesis (TTS) can generate highly natural
synthetic speech from raw text. However, rendering the correct pitch accents is
still a challenging problem for end-to-end TTS. To tackle the challenge of
rendering correct pitch accent in Japanese end-to-end TTS, we adopt PnG~BERT, a
self-supervised pretrained model in the character and phoneme domain for TTS.
We investigate the effects of features captured by PnG~BERT on Japanese TTS by
modifying the fine-tuning condition to determine the conditions helpful
inferring pitch accents. We manipulate content of PnG~BERT features from being
text-oriented to speech-oriented by changing the number of fine-tuned layers
during TTS. In addition, we teach PnG~BERT pitch accent information by
fine-tuning with tone prediction as an additional downstream task. Our
experimental results show that the features of PnG~BERT captured by pretraining
contain information helpful inferring pitch accent, and PnG~BERT outperforms
baseline Tacotron on accent correctness in a listening test. | eess.AS | Investigation of Japanese PnG BERT language model in text-to-speech synthesis for pitch accent language |
2022-12-16T07:11:18Z | http://arxiv.org/pdf/2212.08307v2 | http://arxiv.org/abs/2212.08307v2 | Yuxuan Gu, Xiaocheng Feng, Sicheng Ma, Lingyuan Zhang, Heng Gong, Weihong Zhong, Bing Qin | 2023-05-24T10:02:31Z | cs.CL | Previous work on controllable text generation has explored the idea of
control from the latent space, such as optimizing a representation with
attribute-related classifiers or sampling a representation from relevant
discrete samples. However, they are not effective enough in modeling both the
latent space and the control, leaving controlled text with low quality and
diversity. In this work, we propose a novel control framework using probability
density estimation in the latent space. Our method utilizes an invertible
transformation function, the Normalizing Flow, that maps the complex
distributions in the latent space to simple Gaussian distributions in the prior
space. Thus, we can perform sophisticated and flexible control in the prior
space and feed the control effects back into the latent space owing to the
one-one-mapping property of invertible transformations. Experiments on
single-attribute controls and multi-attribute control reveal that our method
outperforms several strong baselines on attribute relevance and text quality
and achieves the SOTA. Further analysis of control strength adjustment
demonstrates the flexibility of our control strategy. | cs.CL | Controllable Text Generation via Probability Density Estimation in the Latent Space |
2022-12-16T05:17:59Z | http://arxiv.org/pdf/2212.08287v1 | http://arxiv.org/abs/2212.08287v1 | Long Bai, Saiping Guan, Zixuan Li, Jiafeng Guo, Xiaolong Jin, Xueqi Cheng | 2022-12-16T05:17:59Z | cs.CL, cs.AI | Script is a kind of structured knowledge extracted from texts, which contains
a sequence of events. Based on such knowledge, script event prediction aims to
predict the subsequent event. To do so, two aspects should be considered for
events, namely, event description (i.e., what the events should contain) and
event encoding (i.e., how they should be encoded). Most existing methods
describe an event by a verb together with only a few core arguments (i.e.,
subject, object, and indirect object), which are not precise. In addition,
existing event encoders are limited to a fixed number of arguments, which are
not flexible to deal with extra information. Thus, in this paper, we propose
the Rich Event Prediction (REP) framework for script event prediction.
Fundamentally, it is based on the proposed rich event description, which
enriches the existing ones with three kinds of important information, namely,
the senses of verbs, extra semantic roles, and types of participants. REP
contains an event extractor to extract such information from texts. Based on
the extracted rich information, a predictor then selects the most probable
subsequent event. The core component of the predictor is a transformer-based
event encoder to flexibly deal with an arbitrary number of arguments.
Experimental results on the widely used Gigaword Corpus show the effectiveness
of the proposed framework. | cs.CL | Rich Event Modeling for Script Event Prediction |
2022-12-16T05:15:41Z | http://arxiv.org/pdf/2212.08286v2 | http://arxiv.org/abs/2212.08286v2 | Ping Yu, Tianlu Wang, Olga Golovneva, Badr AlKhamissi, Siddharth Verma, Zhijing Jin, Gargi Ghosh, Mona Diab, Asli Celikyilmaz | 2023-07-07T17:43:12Z | cs.CL | Current large language models can perform reasonably well on complex tasks
that require step-by-step reasoning with few-shot learning. Are these models
applying reasoning skills they have learnt during pre-training and reason
outside of their training context, or are they simply memorizing their training
corpus at finer granularity and have learnt to better understand their context?
To tease apart these possibilities, we introduce ALERT, a benchmark and suite
of analyses for assessing language models' reasoning ability comparing
pre-trained and finetuned models on complex tasks that require reasoning skills
to solve. ALERT provides a test bed to asses any language model on fine-grained
reasoning skills, which spans over 20 datasets and covers 10 different
reasoning skills. We leverage ALERT to further investigate the role of
finetuning. With extensive empirical analysis we find that language models
learn more reasoning skills such as textual entailment, abductive reasoning,
and analogical reasoning during finetuning stage compared to pretraining state.
We also find that when language models are finetuned they tend to overfit to
the prompt template, which hurts the robustness of models causing
generalization problems. | cs.CL | ALERT: Adapting Language Models to Reasoning Tasks |
2022-12-16T05:10:09Z | http://arxiv.org/pdf/2212.08283v3 | http://arxiv.org/abs/2212.08283v3 | Feiqi Cao, Siwen Luo, Felipe Nunez, Zean Wen, Josiah Poon, Caren Han | 2023-08-07T08:32:54Z | cs.CV, cs.CL | Most TextVQA approaches focus on the integration of objects, scene texts and
question words by a simple transformer encoder. But this fails to capture the
semantic relations between different modalities. The paper proposes a Scene
Graph based co-Attention Network (SceneGATE) for TextVQA, which reveals the
semantic relations among the objects, Optical Character Recognition (OCR)
tokens and the question words. It is achieved by a TextVQA-based scene graph
that discovers the underlying semantics of an image. We created a
guided-attention module to capture the intra-modal interplay between the
language and the vision as a guidance for inter-modal interactions. To make
explicit teaching of the relations between the two modalities, we proposed and
integrated two attention modules, namely a scene graph-based semantic
relation-aware attention and a positional relation-aware attention. We
conducted extensive experiments on two benchmark datasets, Text-VQA and ST-VQA.
It is shown that our SceneGATE method outperformed existing ones because of the
scene graph and its attention modules. | cs.CV | SceneGATE: Scene-Graph based co-Attention networks for TExt visual question answering |
2022-12-16T04:52:53Z | http://arxiv.org/pdf/2212.08279v1 | http://arxiv.org/abs/2212.08279v1 | Bolin Lai, Hongxin Zhang, Miao Liu, Aryan Pariani, Fiona Ryan, Wenqi Jia, Shirley Anugrah Hayati, James M. Rehg, Diyi Yang | 2022-12-16T04:52:53Z | cs.LG, cs.CL, cs.CV | Persuasion modeling is a key building block for conversational agents.
Existing works in this direction are limited to analyzing textual dialogue
corpus. We argue that visual signals also play an important role in
understanding human persuasive behaviors. In this paper, we introduce the first
multimodal dataset for modeling persuasion behaviors. Our dataset includes 199
dialogue transcriptions and videos captured in a multi-player social deduction
game setting, 26,647 utterance level annotations of persuasion strategy, and
game level annotations of deduction game outcomes. We provide extensive
experiments to show how dialogue context and visual signals benefit persuasion
strategy prediction. We also explore the generalization ability of language
models for persuasion modeling and the role of persuasion strategies in
predicting social deduction game outcomes. Our dataset, code, and models can be
found at https://persuasion-deductiongame.socialai-data.org. | cs.LG | Werewolf Among Us: A Multimodal Dataset for Modeling Persuasion Behaviors in Social Deduction Games |
2022-12-16T03:05:55Z | http://arxiv.org/pdf/2212.08661v1 | http://arxiv.org/abs/2212.08661v1 | Feng Qiu, Chengyang Xie, Yu Ding, Wanzeng Kong | 2022-12-16T03:05:55Z | cs.LG, cs.AI, cs.CL | Humans are skilled in reading the interlocutor's emotion from multimodal
signals, including spoken words, simultaneous speech, and facial expressions.
It is still a challenge to effectively decode emotions from the complex
interactions of multimodal signals. In this paper, we design three kinds of
multimodal latent representations to refine the emotion analysis process and
capture complex multimodal interactions from different views, including a
intact three-modal integrating representation, a modality-shared
representation, and three modality-individual representations. Then, a
modality-semantic hierarchical fusion is proposed to reasonably incorporate
these representations into a comprehensive interaction representation. The
experimental results demonstrate that our EffMulti outperforms the
state-of-the-art methods. The compelling performance benefits from its
well-designed framework with ease of implementation, lower computing
complexity, and less trainable parameters. | cs.LG | EffMulti: Efficiently Modeling Complex Multimodal Interactions for Emotion Analysis |
2022-12-16T01:10:41Z | http://arxiv.org/pdf/2212.08216v2 | http://arxiv.org/abs/2212.08216v2 | Gabrielle Gauthier-Melançon, Orlando Marquez Ayala, Lindsay Brin, Chris Tyler, Frédéric Branchaud-Charron, Joseph Marinier, Karine Grande, Di Le | 2022-12-19T04:01:57Z | cs.LG, cs.AI, cs.CL, cs.HC | We present Azimuth, an open-source and easy-to-use tool to perform error
analysis for text classification. Compared to other stages of the ML
development cycle, such as model training and hyper-parameter tuning, the
process and tooling for the error analysis stage are less mature. However, this
stage is critical for the development of reliable and trustworthy AI systems.
To make error analysis more systematic, we propose an approach comprising
dataset analysis and model quality assessment, which Azimuth facilitates. We
aim to help AI practitioners discover and address areas where the model does
not generalize by leveraging and integrating a range of ML techniques, such as
saliency maps, similarity, uncertainty, and behavioral analyses, all in one
tool. Our code and documentation are available at
github.com/servicenow/azimuth. | cs.LG | Azimuth: Systematic Error Analysis for Text Classification |
2022-12-16T00:21:30Z | http://arxiv.org/pdf/2212.08206v1 | http://arxiv.org/abs/2212.08206v1 | Lakshmi Prasanna Kumar, Arman Kabiri | 2022-12-16T00:21:30Z | cs.CL | Information overloading requires the need for summarizers to extract salient
information from the text. Currently, there is an overload of dialogue data due
to the rise of virtual communication platforms. The rise of Covid-19 has led
people to rely on online communication platforms like Zoom, Slack, Microsoft
Teams, Discord, etc. to conduct their company meetings. Instead of going
through the entire meeting transcripts, people can use meeting summarizers to
select useful data. Nevertheless, there is a lack of comprehensive surveys in
the field of meeting summarizers. In this survey, we aim to cover recent
meeting summarization techniques. Our survey offers a general overview of text
summarization along with datasets and evaluation metrics for meeting
summarization. We also provide the performance of each summarizer on a
leaderboard. We conclude our survey with different challenges in this domain
and potential research opportunities for future researchers. | cs.CL | Meeting Summarization: A Survey of the State of the Art |
2022-12-16T00:15:45Z | http://arxiv.org/pdf/2212.08205v1 | http://arxiv.org/abs/2212.08205v1 | Jiaxuan Li, Richard Futrell | 2022-12-16T00:15:45Z | cs.CL | We advance an information-theoretic model of human language processing in the
brain, in which incoming linguistic input is processed at two levels, in terms
of a heuristic interpretation and in terms of error correction. We propose that
these two kinds of information processing have distinct electroencephalographic
signatures, corresponding to the well-documented N400 and P600 components of
language-related event-related potentials (ERPs). Formally, we show that the
information content (surprisal) of a word in context can be decomposed into two
quantities: (A) heuristic surprise, which signals processing difficulty of word
given its inferred context, and corresponds with the N400 signal; and (B)
discrepancy signal, which reflects divergence between the true context and the
inferred context, and corresponds to the P600 signal. Both of these quantities
can be estimated using modern NLP techniques. We validate our theory by
successfully simulating ERP patterns elicited by a variety of linguistic
manipulations in previously-reported experimental data from Ryskin et al.
(2021). Our theory is in principle compatible with traditional cognitive
theories assuming a `good-enough' heuristic interpretation stage, but with
precise information-theoretic formulation. | cs.CL | A unified information-theoretic model of EEG signatures of human language processing |
2022-12-16T00:15:14Z | http://arxiv.org/pdf/2212.08204v1 | http://arxiv.org/abs/2212.08204v1 | Wenyue Hua, Yuchen Zhang, Zhe Chen, Josie Li, Melanie Weber | 2022-12-16T00:15:14Z | cs.CL, cs.CY | The application of Natural Language Processing (NLP) to specialized domains,
such as the law, has recently received a surge of interest. As many legal
services rely on processing and analyzing large collections of documents,
automating such tasks with NLP tools emerges as a key challenge. Many popular
language models, such as BERT or RoBERTa, are general-purpose models, which
have limitations on processing specialized legal terminology and syntax. In
addition, legal documents may contain specialized vocabulary from other
domains, such as medical terminology in personal injury text. Here, we propose
LegalRelectra, a legal-domain language model that is trained on mixed-domain
legal and medical corpora. We show that our model improves over general-domain
and single-domain medical and legal language models when processing
mixed-domain (personal injury) text. Our training architecture implements the
Electra framework, but utilizes Reformer instead of BERT for its generator and
discriminator. We show that this improves the model's performance on processing
long passages and results in better long-range text comprehension. | cs.CL | LegalRelectra: Mixed-domain Language Modeling for Long-range Legal Text Comprehension |
2022-12-15T23:41:20Z | http://arxiv.org/pdf/2212.08196v1 | http://arxiv.org/abs/2212.08196v1 | Oliver Johnson, Beicheng Lou, Janet Zhong, Andrey Kurenkov | 2022-12-15T23:41:20Z | cs.CL | Often clickbait articles have a title that is phrased as a question or vague
teaser that entices the user to click on the link and read the article to find
the explanation. We developed a system that will automatically find the answer
or explanation of the clickbait hook from the website text so that the user
does not need to read through the text themselves. We fine-tune an extractive
question and answering model (RoBERTa) and an abstractive one (T5), using data
scraped from the 'StopClickbait' Facebook pages and Reddit's 'SavedYouAClick'
subforum. We find that both extractive and abstractive models improve
significantly after finetuning. We find that the extractive model performs
slightly better according to ROUGE scores, while the abstractive one has a
slight edge in terms of BERTscores. | cs.CL | Saved You A Click: Automatically Answering Clickbait Titles |
2022-12-15T23:38:31Z | http://arxiv.org/pdf/2212.08195v1 | http://arxiv.org/abs/2212.08195v1 | Andrew Lee, David Wu, Emily Dinan, Mike Lewis | 2022-12-15T23:38:31Z | cs.CL | Despite many recent advancements in language modeling, state-of-the-art
language models lack grounding in the real world and struggle with tasks
involving complex reasoning. Meanwhile, advances in the symbolic reasoning
capabilities of AI have led to systems that outperform humans in games like
chess and Go (Silver et al., 2018). Chess commentary provides an interesting
domain for bridging these two fields of research, as it requires reasoning over
a complex board state and providing analyses in natural language. In this work
we demonstrate how to combine symbolic reasoning engines with controllable
language models to generate chess commentaries. We conduct experiments to
demonstrate that our approach generates commentaries that are preferred by
human judges over previous baselines. | cs.CL | Improving Chess Commentaries by Combining Language Models with Symbolic Reasoning Engines |