title stringlengths 15 188 | abstract stringlengths 400 1.8k | introduction stringlengths 9 10.5k | content stringlengths 778 41.9k | abstract_len int64 400 1.8k | intro_len int64 9 10.5k | abs_len int64 400 1.8k |
|---|---|---|---|---|---|---|
A Conditional Splitting Framework for Efficient Constituency Parsing | We introduce a generic seq2seq parsing framework that casts constituency parsing problems (syntactic and discourse parsing) into a series of conditional splitting decisions. Our parsing model estimates the conditional probability distribution of possible splitting points in a given text span and supports efficient topd... | A number of formalisms have been introduced to analyze natural language at different linguistic levels. This includes syntactic structures in the form of phrasal and dependency trees, semantic structures in the form of meaning representations In recent years, neural end-to-end parsing methods have outperformed traditio... | Note that in syntactic parsing, the split position must be within the span but not at its edge, that is, k must satisfy i < k < j for each boundary span (i, j). Otherwise, it will not produce valid sub-trees. In this case, we keep splitting until each span contains a single leaf token. However, for discourse trees, eac... | 1,079 | 4,219 | 1,079 |
TripleNet: Triple Attention Network for Multi-Turn Response Selection in Retrieval-based Chatbots | We consider the importance of different utterances in the context for selecting the response usually depends on the current query. 1 In this paper, we propose the model TripleNet to fully model the task with the triple context, query, response instead of context, response in previous works. The heart of TripeNet is a n... | To establish a human-machine dialogue system is one of the most challenging tasks in Artificial Intelligence (AI). Existing works on building dialogue systems are mainly divided into two categories: retrieval-based method A: i downloaded angry ip scanner and now it doesn't work and i can't uninstall it B: you installed... | Earlier works on building the conversation systems are generally based on rules or templates As the portability and coverage of such systems are far from satisfaction, people pay more attention to the data-driven approaches for the opendomain conversation system In this paper, we focus on the task response selection wh... | 829 | 1,448 | 829 |
GL-CLEF: A Global-Local Contrastive Learning Framework for Cross-lingual Spoken Language Understanding | Due to high data demands of current methods, attention to zero-shot cross-lingual spoken language understanding (SLU) has grown, as such approaches greatly reduce human annotation effort. However, existing models solely rely on shared parameters, which can only perform implicit alignment across languages. We present Gl... | Spoken language understanding (SLU) is a critical component in task-oriented dialogue systems [CLS] [CLS] [CLS] To this end, many works have been explored for zero-shot cross-lingual SLU. Multilingual BERT (mBERT) To solve the aforementioned challenges, we propose a Global-Local Contrastive LEarning Framework (GL-CLEF)... | We first describe traditional SLU before the specifics of zero-shot cross-lingual version of SLU. Traditional SLU in Task-oriented Dialogue. SLU in Task-oriented Dialogue contains two subtasks: Intent Detection and Slot Filling. • Intent Detection: Given input utterance x, this is a classification problem to decide the... | 1,139 | 1,598 | 1,139 |
Towards Safer Operations: An Expert-involved Dataset of High-Pressure Gas Incidents for Preventing Future Failures | This paper introduces a new IncidentAI dataset for safety prevention. Different from prior corpora that usually contain a single task, our dataset comprises three tasks: named entity recognition, cause-effect extraction, and information retrieval. The dataset is annotated by domain experts who have at least six years o... | Daily activities usually face incidents that can significantly affect risk management. In specific industries such as manufacturing, an incident can make a significant consequence that not only reduces the reputation of companies but also breaks the product chain and costs a lot of money. It motivates the introduction ... | Incident databases There exist industry-specific incident databases in many industries 3 The HPGIncident Dataset The original dataset was collected from publicly available reports of high-gas incidents published in 2022 by the High-Pressure Gas Safety Institute of Japan. 2 The original data contains descriptions of inc... | 736 | 3,123 | 736 |
Neural semi-Markov CRF for Monolingual Word Alignment | Monolingual word alignment is important for studying fine-grained editing operations (i.e., deletion, addition, and substitution) in textto-text generation tasks, such as paraphrase generation, text simplification, neutralizing biased language, etc. In this paper, we present a novel neural semi-Markov CRF alignment mod... | Monolingual word alignment aims to align words or phrases with similar meaning in two sentences that are written in the same language. It is useful for improving the interpretability in natural language understanding tasks, including semantic textual similarity | Insertion Figure One major challenge for automatic alignment is the need to handle not only alignments between words and linguistic phrases (e.g., a dozen ↔ more than 10), but also non-linguistic phrases that are semantically related given the context (e.g., tensions ↔ relations being strained in Figure Our experimenta... | 980 | 261 | 980 |
WHAT, WHEN, and HOW to Ground: Designing User Persona-Aware Conversational Agents for Engaging Dialogue | This paper presents a method for building a personalized open-domain dialogue system to address the WWH (WHAT, WHEN, and HOW) problem for natural response generation in a commercial setting, where personalized dialogue responses are heavily interleaved with casual response turns. The proposed approach involves weighted... | A personalized dialogue (PD) system is capable of generating user-customized responses based on long-term memory about the user's persona, leading to more trustworthy and engaging conversations The key to enhanced user engagement in a PD system lies in finding a persona that is contextually relevant and appropriate, on... | Since the release of the PersonaChat dataset With respect to the How, However, to the best of our knowledge, there has been no work on addressing all three WWH questions in PD system. Therefore, considering the crucial importance of addressing the WWH issues in a commercial system, we propose novel methods to tackle al... | 863 | 3,171 | 863 |
Thesauruses for Prepositional Phrase Attachment | Probabilistic models have been effective in resolving prepositional phrase attachment ambiguity, but sparse data remains a significant problem. We propose a solution based on similarity-based smoothing, where the probability of new PPs is estimated with information from similar examples generated using a thesaurus. Thr... | Prepositional phrases are an interesting example of syntactic ambiguity and a challenge for automatic parsers. The ambiguity arises whenever a prepositional phrase can modify a preceding verb or noun, as in the canonical example I saw the man with the telescope. In syntactic terms, the prepositional phrase attaches eit... | Early work on PP attachment disambiguation used strictly syntactic or high-level pragmatic rules to decide on an attachment This marked a flowering in the field of PP attachment, with a succession of papers bringing the whole armoury of machine learning techniques to bear on the problem. Smoothing for statistical model... | 845 | 4,220 | 845 |
Gender Inflected or Bias Inflicted: On Using Grammatical Gender Cues for Bias Evaluation in Machine Translation | Neural Machine Translation (NMT) models are state-of-the-art for machine translation. However, these models are known to have various social biases, especially gender bias. Most of the work on evaluating gender bias in NMT has focused primarily on English as the source language. For source languages different from Engl... | Various models trained to learn from data are susceptible to picking up spurious correlations in their training data, which can lead to multiple social biases. In NLP, such biases have been observed in different forms: Even state-of-the-art NMT models develop such biases This problem also exists for HI-EN Machine Trans... | NMT Models : We test HI-EN NMT models which are widely popular and represent state-ofthe-art in both commercial or academic research : (1) IndicTrans Cho et al. ( For translation into English, TGBI uses the fraction of sentences in a sentence set S translated as "masculine", "feminine" or "neutral" in the target , i.e.... | 1,146 | 1,432 | 1,146 |
SDR: Efficient Neural Re-ranking using Succinct Document Representation | BERT based ranking models have achieved superior performance on various information retrieval tasks. However, the large number of parameters and complex self-attention operations come at a significant latency overhead. To remedy this, recent works propose late-interaction architectures, which allow precomputation of in... | Information retrieval (IR) systems traditionally comprise of two stages: retrieval and ranking. Given a user query, the role of the retrieval stage is to quickly retrieve a set of candidate documents * Both authors contributed equally to the paper. † Work carried out while working at Amazon. BERT SPLIT is a distilled l... | Late-interaction models. The idea of running several transformer layers for the document and the query independently, and then combining them in the last transformer layers, was developed concurrently by multiple teams: PreTTR Several other works Compressed embeddings. Our work reduces storage requirements by reducing ... | 567 | 4,277 | 567 |
Understanding and Improving Knowledge Distillation for Quantization-Aware Training of Large Transformer Encoders | Knowledge distillation (KD) has been a ubiquitous method for model compression to strengthen the capability of a lightweight model with the transferred knowledge from the teacher. In particular, KD has been employed in quantization-aware training (QAT) of Transformer encoders like BERT to improve the accuracy of the st... | Knowledge distillation (KD) Quantization-aware training (QAT) stands out for its recent success in reducing not only the memory requirements but also the computational complexity of Transformer models In this work, we provide an in-depth analysis of KD on attention recovery for QAT of Transformers in terms of the knowl... | Transformer-based encoder models like BERT Motivated by where f (x) := (xW V + b V )W O and α i,j is j'th attention probability of i'th token in AM h . Therefore, MHA can be decomposed into two parts: selfattention generation (SA-GEN) corresponding to the attention map (α), and self-attention propagation (SA-PROP) corr... | 1,149 | 2,408 | 1,149 |
Discovering Differences in the Representation of People using Contextualized Semantic Axes | A common paradigm for identifying semantic differences across social and temporal contexts is the use of static word embeddings and their distances. In particular, past work has compared embeddings against "semantic axes" that represent two opposing concepts. We extend this paradigm to BERT embeddings, and construct co... | Warning: This paper contains content that may be offensive or upsetting. Quantifying and describing the nature of language differences is key to measuring the impact of social and cultural factors on text. Past work has compared English embeddings for people to adjectives or concepts embeddings of less frequent minorit... | Static embeddings. Several formulae for calculating the similarity of a target word to two sets of pole words have been proposed in prior work on static semantic axes. These differ in whether they take the difference between a target word's similarities to each pole Relying on single-word poles for axes can be unstable... | 831 | 1,395 | 831 |
Scalable Term Selection for Text Categorization | In text categorization, term selection is an important step for the sake of both categorization accuracy and computational efficiency. Different dimensionalities are expected under different practical resource restrictions of time or space. Traditionally in text categorization, the same scoring or ranking criterion is ... | Text categorization is a classical text information processing task which has been studied adequately • Many irrelevant terms have detrimental effect on categorization accuracy due to overfitting • Some text categorization tasks have many relevant but redundant features, which also hurt the categorization accuracy (i) ... | Two document collections are used in this study. CE (Chinese Encyclopedia): This is from the electronic version of the Chinese Encyclopedia. We choose a Chinese corpus as the primary document collection because Chinese text (as well as other Asian languages) has a very large term set and a satisfying subset is usually ... | 1,237 | 3,115 | 1,237 |
Learning Language and Multimodal Privacy-Preserving Markers of Mood from Mobile Data | Mental health conditions remain underdiagnosed even in countries with common access to advanced medical care. The ability to accurately and efficiently predict mood from easily collectible data has several important implications for the early detection, intervention, and treatment of mental health disorders. One promis... | Mental illnesses can have a damaging permanent impact on communities, societies, and economies all over the world | Figure Intensive monitoring of behaviors via adolescents' natural use of smartphones may help identify realtime predictors of mood in high-risk youth as a proxy for suicide risk Recent work in affective computing has begun to explore the potential in predicting mood from mobile data. Studies have found that typing patt... | 1,319 | 113 | 1,319 |
Dating Greek Papyri with Text Regression | Dating Greek papyri accurately is crucial not only to edit their texts but also to understand numerous other aspects of ancient writing, document and book production and circulation, as well as various other aspects of administration, everyday life and intellectual history of antiquity. Although a substantial number of... | Ancient textual artefacts are arguably the richest source of information on the ancient world. In the Graeco-Roman world and particularly in its Greekspeaking part, the most extensive coeval texts come from inscriptions and papyri. The latter is a collective term used for all ancient manuscripts, regardless of their wr... | Dating historical documents with computational means has been studied for many languages The studied languages are Latin The employed methods usually were standard machine learning methods, such as KNN Pre-trained convolutional neural networks have been used to extract features, which are passed to a classifier or regr... | 870 | 3,639 | 870 |
A Fast Boosting-based Learner for Feature-Rich Tagging and Chunking | Combination of features contributes to a significant improvement in accuracy on tasks such as part-of-speech (POS) tagging and text chunking, compared with using atomic features. However, selecting combination of features on learning with large-scale and feature-rich training data requires long training time. We propos... | Several boosting-based learning algorithms have been applied to Natural Language Processing problems successfully. These include text categorization Furthermore, classifiers based on boostingbased learners have shown fast classification speed However, boosting-based learning algorithms require long training time. One o... | We describe the problem treated by our boostingbased learner as follows. Let X be the set of examples and Y be a set of labels {-1, +1}. Let F = {f 1 , f 2 , ..., f M } be M types of features represented by strings. Let S be a set of training sam-## S = {(xi, yi)} m i=1 : xi ⊆ X ,yi ∈ {±1} ## a smoothing value ε =1 ## ... | 1,155 | 1,866 | 1,155 |
Cree Corpus: A Collection of nêhiyawêwin Resources | Plains Cree (nêhiyawêwin) is an Indigenous language that is spoken in Canada and the USA. It is the most widely spoken dialect of Cree and a morphologically complex language that is polysynthetic, highly inflective, and agglutinative. It is an extremely low resource language, with no existing corpus that is both availa... | Recent work with Indigenous persons has shown that some want advanced technologies to support the learning and use of their languages. The Cree and Métis persons involved in this study stated a desire for technologies such as an app to help with learning the structure of the language for conversation, translation, and ... | Plains Cree is called nêhiyawêwin by its speakers, and it is not capitalized. nêhiyawêwin is a widely-spoken dialect of the Indigenous language that English-speakers call Cree: nêhiyawêwin is the mother tongue for approximately 3,655 speakers, and it is the language spoken most at home for approximately 2,165 persons C... | 1,000 | 2,087 | 1,000 |
Exploring Dynamic Selection of Branch Expansion Orders for Code Generation | Due to the great potential in facilitating software development, code generation has attracted increasing attention recently. Generally, dominant models are Seq2Tree models, which convert the input natural language description into a sequence of tree-construction actions corresponding to the pre-order traversal of an A... | Code generation aims at automatically generating a source code snippet given a natural language (NL) description, which has attracted increasing attention recently due to its potential value in simplifying programming. Instead of modeling the abstract syntax tree (AST) of code snippets directly, most of methods for cod... | Only TRANX 8.47 Only TRANX-R2L 7.66 Table ones. Table In this paper, we explore dynamic selection of branch expansion orders for code generation. Specifically, we propose to equip the conventional Seq2Tree model with a context-based Branch Selector, which dynamically quantifies the priorities of expanding different bra... | 1,104 | 863 | 1,104 |
Learning Prototypical Goal Activities for Locations | People go to different places to engage in activities that reflect their goals. For example, people go to restaurants to eat, libraries to study, and churches to pray. We refer to an activity that represents a common reason why people typically go to a location as a prototypical goal activity (goal-act). Our research a... | Every day, people go to different places to accomplish goals. People go to stores to buy clothing, go to restaurants to eat, and go to the doctor for medical services. People travel to specific destinations to enjoy the beach, go skiing, or see historical sites. For most places, people typically go there for a common s... | Recognizing plans and goals is fundamental to narrative story understanding Goals and plans can also function to trigger scripts Graph-based semi-supervised learning has been successfully used for many tasks, including sentiment analysis Our aim is to learn the most prototypical goal-acts for locations. To tackle this ... | 838 | 2,826 | 838 |
Sentence-Level Agreement for Neural Machine Translation | The training objective of neural machine translation (NMT) is to minimize the loss between the words in the translated sentences and those in the references. In NMT, there is a natural correspondence between the source sentence and the target sentence. However, this relationship has only been represented using the enti... | Neural network based methods have been applied to several natural language processing tasks Based on this hypothesis, Sentence-level agreement method has been applied to many natural language processing tasks. Aliguliyev (2009) used sentence similarity measure technique for automatic text summarization. In human transl... | In this section, we take the Transformer architecture proposed by As an encoder-to-decoder architecture, X = {x 1 , x 2 , ..., x J } represents a source sentence and Y = {y 1 , y 2 , ..., y I } represents a target sentence. The encoder-to-decoder model learns to estimate the conditional probability from the source sent... | 884 | 1,169 | 884 |
Factored Translation Models | We present an extension of phrase-based statistical machine translation models that enables the straight-forward integration of additional annotation at the word-levelmay it be linguistic markup or automatically generated word classes. In a number of experiments we show that factored translation models lead to better t... | The current state-of-the-art approach to statistical machine translation, so-called phrase-based models, is limited to the mapping of small text chunks without any explicit use of linguistic information, may it be morphological, syntactic, or semantic. Such additional information has been demonstrated to be valuable by... | Many attempts have been made to add richer information to statistical machine translation models. Most of these focus on the pre-processing of the input to the statistical system, or the post-processing of its output. Our framework is more general and goes beyond recent work on models that back off to representations w... | 417 | 1,936 | 417 |
Finding Syntax in Human Encephalography with Beam Search | Recurrent neural network grammars (RNNGs) are generative models of (tree, string) pairs that rely on neural networks to evaluate derivational choices. Parsing with them using beam search yields a variety of incremental complexity metrics such as word surprisal and parser action count. When used as regressors against hu... | Computational psycholinguistics has "always been...the thing that computational linguistics stood the greatest chance of providing to humanity" As The contribution of the present paper is situated precisely at this intersection. It combines a probabilistic generative grammar (RNNG; Comparison with language models based... | Recurrent neural network grammars (henceforth: RNNGs Figure Each step of this generative story depends on the state of a stack, depicted inside the gray box in Figure Phrase-closing actions trigger a syntactic composition function (depicted in Figure The parameters of all these components are adaptively adjusted using ... | 453 | 1,204 | 453 |
Modeling Dual Read/Write Paths for Simultaneous Machine Translation | Simultaneous machine translation (SiMT) outputs translation while reading source sentence and hence requires a policy to decide whether to wait for the next source word (READ) or generate a target word (WRITE), the actions of which form a read/write path. Although the read/write path is essential to SiMT performance, n... | Simultaneous machine translation (SiMT) The sequence of READ and WRITE actions in the translation process form a read/write path, which is key to SiMT performance. Improper read/write path will bring damage to translation performance as compared to the following WRITE actions too many but not necessary READ actions wi... | We first briefly introduce SiMT with a focus on monotonic multi-head attention For a SiMT task, we denote the source sentence as Read/write path can be represented in multiple forms, such as an action sequence of READ and WRITE (e.g., RRWWWRW• • • ), or a path from (0, 0) to (I, J) in the attention matrix from the targ... | 867 | 2,669 | 867 |
Learning to Jointly Predict Ellipsis and Comparison Structures | Domain-independent meaning representation of text has received a renewed interest in the NLP community. Comparison plays a crucial role in shaping objective and subjective opinion and measurement in natural language, and is often expressed in complex constructions including ellipsis. In this paper, we introduce a novel... | Representing the underlying meaning of text has been a long-standing topic of interest in computational linguistics. Recently there has been a renewed interest in representation of meaning for various tasks such as semantic parsing, where the task is to map a natural language sentence into its corresponding formal mean... | Broadly, elliptical constructions involve the omission of one or more phrases from a clause (such as 'drove fast' phrase at the end of example (1)) whose content can still be fully recovered from the unelided words of the sentence In 2010, a SemEval task was organized with the goals of (1) automatically detecting VPE i... | 829 | 2,508 | 829 |
Deconfounding Legal Judgment Prediction for European Court of Human Rights Cases Towards Better Alignment with Experts | This work demonstrates that Legal Judgement Prediction systems without expert-informed adjustments can be vulnerable to shallow, distracting surface signals that arise from corpus construction, case distribution, and confounding factors. To mitigate this, we use domain expertise to strategically identify statistically ... | The task of Legal Judgment Prediction (LJP) has recently gained increasing attention in the legal and mainstream NLP communities In this work, we focus on LJP for the European Court of Human Rights (ECtHR), which adjudicates complaints by individuals against states about alleged violations of their rights as enshrined ... | LJP as an NLP task has been tackled using ngram representations (e.g., The ECtHR has been the subject of substantial prior work in LJP. We use two datasets for model training and evaluation: First, for binary violation we use the dataset by We conduct experiments on four LJP tasks: Task J -Binary Violation For our task... | 904 | 2,492 | 904 |
RNG-KBQA: Generation Augmented Iterative Ranking for Knowledge Base Question Answering | Existing KBQA approaches, despite achieving strong performance on i.i.d. test data, often struggle in generalizing to questions involving unseen KB schema items. Prior rankingbased approaches have shown some success in generalization, but suffer from the coverage issue. We present RnG-KBQA, a Rank-and-Generate approach... | Modern knowledge bases (KB) are reliable sources of a huge amount of world knowledge but can be difficult to interact with since they are extremely large in scale and require specific query languages (e.g., Sparql) to access. Question Answering over Knowledge Base (KBQA) serves as a user-friendly way to query over KBs ... | A knowledge base collects knowledge data stored in the form of subject-relation-object triple (s, r, o), where s is an entity, r is a binary relation, and o can be entities or literals (e.g., date time, integer values, etc.). Let the question be x, our task is to obtain a logical form y that can be executed over the kn... | 1,373 | 1,394 | 1,373 |
Rich bitext projection features for parse reranking | Many different types of features have been shown to improve accuracy in parse reranking. A class of features that thus far has not been considered is based on a projection of the syntactic structure of a translation of the text to be parsed. The intuition for using this type of bitext projection feature is that ambiguo... | Parallel text or bitext is an important knowledge source for solving many problems such as machine translation, cross-language information retrieval, and the projection of linguistic resources from one language to another. In this paper, we show that bitext-based features are effective in addressing another NLP problem... | Consider the English sentence "He saw a baby and a woman who had gray hair". Suppose that the baseline parser generates two parses, containing the NPs shown in figures 1 and 2, respectively, and that the semantically more plausible second parse in figure 2 is correct. How can we determine that the second parse should b... | 505 | 2,012 | 505 |
Transforming Meaning Representation Grammars to Improve Semantic Parsing | A semantic parser learning system learns to map natural language sentences into their domain-specific formal meaning representations, but if the constructs of the meaning representation language do not correspond well with the natural language then the system may not learn a good semantic parser. This paper presents ap... | Semantic parsing is the task of converting natural language (NL) sentences into their meaning representations (MRs) which a computer program can execute to perform some domain-specific task, like controlling a robot, answering database queries etc. These MRs are expressed in a formal meaning representation language (MR... | The following subsection gives some examples of semantic parsing domains and their corresponding MRLs and illustrates why incompatibility between MRGs and natural language could hurt semantic parsing. The next subsection then briefly describes a base semantic parser learning system which we use in our experiments. ativ... | 846 | 966 | 846 |
Learning to Generate Equitable Text in Dialogue from Biased Training Data | The ingrained principles of fairness in a dialogue system's decision-making process and generated responses are crucial for user engagement, satisfaction, and task achievement. Absence of equitable and inclusive principles can hinder the formation of common ground, which in turn negatively impacts the overall performan... | Machine learning models for text-generation in dialogue have trouble learning the "long tail" of a data distribution; i.e., the data concepts not frequently observed during training. For example, dataset biases like gender imbalance can induce a long tail in training data whereby important data relationships involving ... | Recent proposals for the use of learning theory in dialogue are due to (1) Of course, there are a number of undefined terms here: specifically, the test h, the context C, the goal dialogue D, the learned dialogue D, and the unobserved effects U . Below, we explain each, using examples from Figure The goal distribution ... | 1,398 | 1,776 | 1,398 |
Word Embeddings through Hellinger PCA | Word embeddings resulting from neural language models have been shown to be a great asset for a large variety of NLP tasks. However, such architecture might be difficult and time-consuming to train. Instead, we propose to drastically simplify the word embeddings computation through a Hellinger PCA of the word cooccuren... | Building word embeddings has always generated much interest for linguists. Popular approaches such as Brown clustering algorithm This paper aims to show that such good word embeddings can be obtained using simple (mostly linear) operations. We show that similar word embeddings can be computed using the word cooccurrenc... | As 80% of the meaning of English text comes from word choice and the remaining 20% comes from word order Linguists assumed long ago that words occurring in similar contexts tend to have similar meanings It has been shown that using word embeddings as features helps to improve general performance on many NLP tasks A NNL... | 676 | 1,027 | 676 |
Attention and Edge-Label Guided Graph Convolutional Networks for Named Entity Recognition | It has been shown that named entity recognition (NER) could benefit from incorporating the long-distance structured information captured by dependency trees. However, dependency trees built by tools usually have a certain percentage of errors. Under such circumstances, how to better use relevant structured information ... | Named Entity Recognition (NER) is the recognition of entities with specific meanings in the text, mainly including person, organization, location, etc. NER is the fundamental tasks for many natural language processing tasks such as relation extraction A dependency tree reveals the syntactic structure of a language unit... | The traditional feature-based NER approaches require considerable feature engineering skills and domain expertise. However, deep neural network based models can build reliable NER systems with much less effort in designing features. BiLSTM-CRF model To further improve named entity recognition, the representation of wor... | 1,317 | 2,787 | 1,317 |
Marine Variable Linker: Exploring Relations between Changing Variables in Marine Science Literature | We report on a demonstration system for text mining of literature in marine science and related disciplines. It automatically extracts variables (e.g. CO2) involved in events of change/increase/decrease (e.g increasing CO2), as well as cooccurrence and causal relations among these events (e.g. increasing CO2 causes a d... | Progress in science relies significantly on the premise that -in addition to other methods for gaining knowledge such as experiments and modelling -new knowledge can be inferred by combining existing knowledge found in the literature. Unfortunately such knowledge often remains undiscovered because individual researcher... | Although graph search queries can be written by hand, it takes time, effort and a considerable amount of expertise. In addition, large tables are difficult to read and navigate, lacking an easy way to browse the results, e.g., to look up the source sentences and articles for extracted events. Moreover, users need to ha... | 548 | 4,876 | 548 |
Backward Compatibility During Data Updates by Weight Interpolation | Backward compatibility of model predictions is a desired property when updating a machine learning driven application. It allows to seamlessly improve the underlying model without introducing regression bugs. In classification tasks these bugs occur in the form of negative flips. This means an instance that was correct... | In conventional software development it is an established routine to identify and fix regression bugs before deploying a new version. Regression bugs describe defects in already existing features and are particularly sensible for end users because accustomed workflows are affected. In machine learning driven applicatio... | Old Model New Model Figure mitigated. This prevents backward compatibility of e.g. visual search systems To reduce negative flips during data updates, we propose Backward Compatible Weight Interpolation (BCWI) in this paper. BCWI describes the interpolation between the weights of the old model and the weights of the ne... | 1,296 | 491 | 1,296 |
What's The Latest? A Question-driven News Chatbot | This work describes an automatic news chatbot that draws content from a diverse set of news articles and creates conversations with a user about the news. Key components of the system include the automatic organization of news articles into topical chatrooms, integration of automatically generated questions into the co... | Chatbots offer the ability for interactive information access, which could be of great value in the news domain. As a user reads through news content, interaction could enable them to ask clarifying questions and go in depth on selected subjects. Current news chatbots have minimal capabilities, with content hand-crafte... | This section describes the components of the chatbot: the content source, the user interface, the supported user actions and the computed system answers. Appendix A lists library and data resources used in the system. We form the content for the chatbot from a set of news sources. We have collected an average of 2,000 ... | 633 | 1,834 | 633 |
GWLAN: General Word-Level AutocompletioN for Computer-Aided Translation | Computer-aided translation (CAT), the use of software to assist a human translator in the translation process, has been proven to be useful in enhancing the productivity of human translators. Autocompletion, which suggests translation results according to the text pieces provided by human translators, is a core functio... | Machine translation (MT) has witnessed great advancements with the emergence of neural machine translation (NMT) We asked two sp Wir haben die Meinung von zwei Fachärzten eingeholt. We asked two experts for their opinion. We sp their opinion. 2009). In spite of this, MT systems cannot replace human translators, especia... | Computer-aided translation (CAT) is a widely used practice when using MT technology in the industry. As the the MT systems advanced and improved, various efficient interaction ways of CAT have emerged Sentence-level Autocompletion Most of previous work in autocompletion for CAT focus on sentence-level completion. A com... | 1,235 | 2,503 | 1,235 |
Speculative Contrastive Decoding | Large language models (LLMs) exhibit exceptional performance in language tasks, yet their auto-regressive inference is limited due to high computational requirements and is suboptimal due to the exposure bias. Inspired by speculative decoding and contrastive decoding, we introduce Speculative Contrastive Decoding (SCD)... | Large language models (LLMs) have advanced the versatility and proficiency in approaching realworld natural language tasks such as general instruction following As for decoding acceleration, one prominent method named speculative decoding As for the generation quality, contrastive decoding has been recently proposed In... | In terms of inference acceleration, recent research has been devoted to developing various efficient decoding methods In terms of inference quality, rich research has been suggested We follow the terminology in The intrinsic rationale of contrastive decoding (CD) is that amateur LMs have stronger systematic undesirable... | 688 | 1,146 | 688 |
Classification and Clustering of Arguments with Contextualized Word Embeddings | We experiment with two recent contextualized word embedding methods (ELMo and BERT) in the context of open-domain argument search. For the first time, we show how to leverage the power of contextualized word embeddings to classify and cluster topic-dependent arguments, achieving impressive results on both tasks and acr... | Argument mining methods have been applied to different tasks such as identifying reasoning structures Identifying arguments for unseen topics is a challenging task for machine learning systems. The lexical appearance for two topics, e.g. "net neutrality" and "school uniforms", is vastly different. Hence, in order to pe... | In the following, we concentrate on the fundamental tasks involved in open-domain argument search. First, we discuss work that experiments with sentence-level argument classification. Second, we review work that provides us with the necessary tools to cluster extracted arguments by their similarity. Third, we take a de... | 796 | 1,849 | 796 |
Temporally-Informed Analysis of Named Entity Recognition | Natural language processing models often have to make predictions on text data that evolves over time as a result of changes in language use or the information described in the text. However, evaluation results on existing data sets are seldom reported by taking the timestamp of the document into account. We analyze an... | Natural language processing models are now deployed on a large scale in many applications and used to drive automatic analyses or for making predictions. The usual setup is that these models are trained and evaluated on the data available at model building time, but are used to make inferences on data coming in at a fu... | Language change is a popular topic of research in linguistics Temporal information has been used to create topic models of better quality, usually by adding smoothing properties Most similar to our experimental setup, A related, but distinct, task built on the assumption of language change with time is automatic predic... | 983 | 1,508 | 983 |
Annotation and Automatic Classification of Aspectual Categories | We present the first annotated resource for the aspectual classification of German verb tokens in their clausal context. We use aspectual features compatible with the plurality of aspectual classifications in previous work and treat aspectual ambiguity systematically. We evaluate our corpus by using it to train supervi... | The universal linguistic category of aspect describes how a verb or a verbal projection (including sentences, 'predicates' for short) characterises the temporal course of a state of affairs or 'eventuality'. Such information is relevant for tasks that extract temporal information from texts, such as information extract... | Aspectual classes are established by feature dichotomies Dynamic predicates can be either unbounded (introduce eventualities without inherent boundaries, e.g., move or play the piano), or bounded (e.g., run a mile or build a house). Bounded predicates (also called 'telic') have four subgroups that are crossclassified b... | 453 | 772 | 453 |
Lazy-k: Decoding for Constrained Information Extraction | We explore the possibility of improving probabilistic models in structured prediction. Specifically, we combine the models with constrained decoding approaches in the context of token classification for information extraction. The decoding methods search for constraintsatisfying label-assignments while maximizing the t... | Much of today's Information Extraction (IE) is done using probability-based token-classification models such as BERT Ideally, alternative, high-likelihood predictions are explored to improve predictions from existing models. This is especially interesting in structuredprediction tasks, where the model's predictions are... | To search over high-probability predictions, we require a probabilistic model that outputs independent probabilities for a given sequence of tokens. Given an input sequence x = {x 1 , x 2 , . . . , x n }, x i ∈ X where X is the token vocabulary, the goal is to estimate the probability of the output sequence y = {y 1 , ... | 737 | 2,284 | 737 |
AMesure: a web platform to assist the clear writing of administrative texts | This article presents the AMesure platform, which aims to assist writers of French administrative texts in simplifying their writing. This platform includes a readability formula specialized for administrative texts and it also uses various natural language processing (NLP) tools to analyze texts and highlight a number... | In our current society, written documents play a central role as an information channel, especially in the context of communication between institutions and their target audiences Administrations have been aware of this issue for decades and have launched various initiatives to address it, the most prominent of which i... | This work stands at the intersection between two very different fields: writing studies -"the interdisciplinary science that studies all the processes and knowledge involved in the production of professional writings and their appropriateness for the addressees" Relevant facts from writing studies have already been cov... | 608 | 838 | 608 |
BabyStories: Can Reinforcement Learning Teach Baby Language Models to Write Better Stories? | Language models have seen significant growth in the size of their corpus, leading to notable performance improvements. Yet, there has been limited progress in developing models that handle smaller, more human-like datasets. As part of the BabyLM shared task, this study explores the impact of reinforcement learning from... | The recent growth in the size of large language models (LLMs) has enhanced natural language processing capabilities, from information extraction Storytelling is a fundamental human activity used to share information, impart lessons, and keep loved ones informed about our daily lives The performance of small language mo... | Research has shown that smaller models tend to underperform when trained on large datasets, making the study of model downscaling a non-trivial Our research, however, is driven by a desire to understand if small pretrained models can benefit from Reinforcement Learning from Human Feedback (RLHF), potentially improving ... | 1,009 | 910 | 1,009 |
Exploiting Multi-Word Units in History-Based Probabilistic Generation | We present a simple history-based model for sentence generation from LFG f-structures, which improves on the accuracy of previous models by breaking down PCFG independence assumptions so that more f-structure conditioning context is used in the prediction of grammar rule expansions. In addition, we present work on expe... | Sentence generation, or surface realisation, is the task of generating meaningful, grammatically correct and fluent text from some abstract semantic or syntactic representation of the sentence. It is an important and growing field of natural language processing with applications in areas such as transferbased machine t... | In (statistical) generators, sentences are generated from an abstract linguistic encoding via the application of grammar rules. These rules can be handcrafted grammar rules, such as those of Insofar as it is a broad coverage generator, which has been trained and tested on sections of the WSJ corpus, our generator is cl... | 626 | 1,203 | 626 |
LVP-M 3 : Language-aware Visual Prompt for Multilingual Multimodal Machine Translation | Multimodal Machine Translation (MMT) focuses on enhancing text-only translation with visual features, which has attracted considerable attention from both natural language processing and computer vision communities. Recent advances still struggle to train a separate model for each language pair, which is costly and una... | Multimodal Machine Translation (MMT) extends the conventional text-based machine translation by taking corresponding images as additional inputs However, as shown in Fig. To eliminate the above limitations, we propose a simple and effective LVP-M 3 method, including Token Encoding, Language-aware Visual Prompt Generati... | Multimodal Machine Translation. The multimodal context plays a key role in Multimodal Machine Translation (MMT). Recent MMT methods can be divided into three categories: (1) Using global visual features directly (3) Combining other vision tasks with the translation task by multitask learning Vision-Language Models. The... | 1,248 | 1,558 | 1,248 |
The economic trade-offs of large language models: A case study | Contacting customer service via chat is a common practice. Because employing customer service agents is expensive, many companies are turning to NLP that assists human agents by auto-generating responses that can be used directly or with modifications. Large Language Models (LLMs) are a natural fit for this use case; h... | Amidst increased automation, human agents continue to play an important role in providing excellent customer service. While many conversations are automated in text-based customer support, others are routed to human agents who can handle certain customer concerns more effectively. Agents often handle multiple conversat... | Transformers The size of these LLMs plays a significant role in their high performance Response generation is difficult to evaluate holistically. Some have focused on relevance and level of detail We calculate the ENCS for each model using equation (2), repeated here in (5). ( Using the RU scores in Tables The factor w... | 1,101 | 1,442 | 1,101 |
From Raw Text to Universal Dependencies -Look, No Tags! | We present the Uppsala submission to the CoNLL 2017 shared task on parsing from raw text to universal dependencies. Our system is a simple pipeline consisting of two components. The first performs joint word and sentence segmentation on raw text; the second predicts dependency trees from raw words. The parser bypasses ... | The CoNLL 2017 shared task differs from most previous multilingual dependency parsing tasks not only by using cross-linguistically consistent syntactic representations from the UD project The Uppsala team has adopted a minimalistic stance in this respect and developed a system that does not predict any linguistic struc... | We model joint sentence and word segmentation as a character-level sequence labeling problem in a Bi-RNN-CRF model In the BiRNN-CRF architecture, charactersregardless of writing system -are represented as dense vectors and fed into the bidirectional recurrent layers. We employ the gated recurrent unit (GRU) As illustra... | 611 | 2,101 | 611 |
FlowEval: A Consensus-Based Dialogue Evaluation Framework Using Segment Act Flows | Despite recent progress in open-domain dialogue evaluation, how to develop automatic metrics remains an open problem. We explore the potential of dialogue evaluation featuring dialog act information, which was hardly explicitly modeled in previous methods. However, defined at the utterance level in general, dialog act ... | Dialogue evaluation plays a crucial role in the recent advancement of dialogue research. While human evaluation is often considered as a universal and reliable method by the community Traditional word-overlap metrics, like BLEU | Segment Act Flow Speaker1: How are you? May I have a cup of coffee? greeting , directive Speaker2: Hmm. Certainly. What kind of coffee do you like? We have espresso and latte. backchannelsuccess , commissive , question , inform Table 2021) and harnessing the power of large models One difficulty of using segment act for... | 1,155 | 227 | 1,155 |
Look Harder: A Neural Machine Translation Model with Hard Attention | Soft-attention based Neural Machine Translation (NMT) models have achieved promising results on several translation tasks. These models attend all the words in the source sequence for each target token, which makes them ineffective for long sequence translation. In this work, we propose a hard-attention based NMT model... | In recent years, soft-attention based neural machine translation models Different attention mechanisms have been proposed to improve the quality of the context vector. For example, To overcome the shortcomings of the above approaches, we propose a hard-attention mechanism for a deep NMT model | A typical NMT model based on encoder-decoder architecture generates a target sequence y = {y 1 , • • • , y n } given a source sequence x = {x 1 , • • • , x m } by modeling the conditional probability p(y|x, θ). The encoder (θ e ) computes a set of representations Z = {z 1 , • • • , z m } ∈ R m×d corresponding to x and ... | 851 | 293 | 851 |
IBADR: an Iterative Bias-Aware Dataset Refinement Framework for Debiasing NLU models | As commonly-used methods for debiasing natural language understanding (NLU) models, dataset refinement approaches heavily rely on manual data analysis, and thus maybe unable to cover all the potential biased features. In this paper, we propose IBADR, an Iterative Bias-Aware Dataset Refinement framework, which debiases ... | Although neural models have made significant progress in many natural language understanding (NLU) tasks To alleviate this issue, researchers have proposed many methods that can be generally divided into two categories: model-centric mitigation approaches In this paper, we propose IBADR, an Iterative Bias-Aware Dataset... | In this section, we give a detailed description of IBADR. Under this framework, we first use a limited set of training samples to train a shallow model, which serves to measure the bias degree of samples. Then, we iteratively generate pseudo samples with fewer biased features, as illustrated in Figure As investigated i... | 1,204 | 2,415 | 1,204 |
SoNLP-DP System for ConLL-2016 English Shallow Discourse Parsing | This paper describes the submitted English shallow discourse parsing system from the natural language processing (NLP) group of Soochow university (SoNLP-DP) to the CoNLL-2016 shared task. Our System classifies discourse relations into explicit and non-explicit relations and uses a pipeline platform to conduct every su... | Discourse parsing determines the internal structure of a text via identifying the discourse relations between its text units and plays an important role in natural language understanding that benefits a wide range of downstream natural language applications, such as coherence modeling As the largest discourse corpus, t... | In this section, after a quick overview of our system, we describe the details involved in implementing the end-to-end shallow discourse parser. A typical text consists of sentences glued together in a systematic way to form a coherent discourse. Referring to the PDTB, shallow discourse parsing focus on shallow discour... | 586 | 1,685 | 586 |
Modeling Legal Reasoning: LM Annotation at the Edge of Human Agreement | Generative language models (LMs) are increasingly used for document class-prediction tasks and promise enormous improvements in cost and efficiency. Existing research often examines simple classification tasks, but the capability of LMs to classify on complex or specialized tasks is less well understood. We consider a ... | Academia and industry increasingly use generative language models (LMs) for document annotation and class-prediction tasks, which promise enormous improvements in cost and efficiency. However, research tends to focus on relatively simple and generic annotation contexts, such as topic or query-keyword relevance In this ... | Researchers have developed strategies to guide LMs to perform complex tasks without the time and infrastructure costs of fine-tuning, often by breaking decisions down into multiple steps of reasoning. At certain tasks and with these prompting strategies, LMs perform annotation or classification tasks at the level of hu... | 1,364 | 3,021 | 1,364 |
Multi-Task Pre-Training for Plug-and-Play Task-Oriented Dialogue System | Pre-trained language models have been recently shown to benefit task-oriented dialogue (TOD) systems. Despite their success, existing methods often formulate this task as a cascaded generation problem which can lead to error accumulation across different sub-tasks and greater data annotation overhead. In this study, we... | Task-oriented dialogue is often decomposed into three sub-tasks: (1) dialogue state tracking (DST) for tracking user's belief state; (2) dialogue policy learning (POL) for deciding which system action to take; (3) natural language generation (NLG) for generating dialogue response Traditional approaches With the advance... | Task-Oriented Dialogue. Task-oriented dialogue aims at accomplishing user's goal. Traditional systems Language Model Pre-training. The research community has witnessed remarkable progress of pre-training methods in a wide range of NLP tasks, including language understanding In the dialogue domain, many models are pretr... | 1,047 | 1,574 | 1,047 |
Don't Let Discourse Confine Your Model: Sequence Perturbations for Improved Event Language Models | Event language models represent plausible sequences of events. Most existing approaches train autoregressive models on text, which successfully capture event co-occurrence but unfortunately constrain the model to follow the discourse order in which events are presented. Other domains may employ different discourse orde... | Event-level language models (LMs) provide a way to reason about events, and to approximate schematic and script-like knowledge In this paper, we aim to improve event-level LMs in order to make them more suitable for general knowledge learning. While a range of possible modifications to the model can be imagined, such a... | Event language modeling tasks are typically defined over sequences of events as they appear in text. The events can be represented either as a sequence of words annotated with predicateargument structure (e.g., semantic roles However, relying on discourse order may not be necessary and can potentially limit generalizat... | 861 | 674 | 861 |
Training with Adversaries to Improve Faithfulness of Attention in Neural Machine Translation | Can we trust that the attention heatmaps produced by a neural machine translation (NMT) model reflect its true internal reasoning? We isolate and examine in detail the notion of faithfulness in NMT models. We provide a measure of faithfulness for NMT based on a variety of stress tests where model parameters are perturb... | Can we trust our neural models? This question has led to a wide variety of contemporary NLP research focusing on (a) different axes of interpretability including plausibility (or interchangeably human-interpretability) Aligned with these criteria, we study faithfulness of attention in NMT, the extent to which it can re... | Intuitively, a faithful explanation should reflect the true internal reasoning of the model. Although there is no formal definition for faithfulness, a common approach in the community is to design stress tests to perturb the model parameters chosen in such a way that the model's decision should change if the model is ... | 1,027 | 1,185 | 1,027 |
Dependency-driven Relation Extraction with Attentive Graph Convolutional Networks | Syntactic information, especially dependency trees, has been widely used by existing studies to improve relation extraction with better semantic guidance for analyzing the context information associated with the given entities. However, most existing studies suffer from the noise in the dependency trees, especially whe... | Relation extraction (RE), which aims to detect the relationship between entity mentions from raw text, is one of the most important tasks in information extraction and retrieval, and plays a crucial role in supporting many downstream natural language processing (NLP) applications such as text mining et al., 2019), ques... | RE is conventionally performed as a typical classification task. Our approach follows this paradigm by using A-GCN and incorporates dependency information to improve model performance, where the overall architecture of our model is illustrated in Figure where T X is the dependency tree of X obtained from an off-the-she... | 1,294 | 1,590 | 1,294 |
DualSum: a Topic-Model based approach for update summarization | Update summarization is a new challenge in multi-document summarization focusing on summarizing a set of recent documents relatively to another set of earlier documents. We present an unsupervised probabilistic approach to model novelty in a document collection and apply it to the generation of update summaries. The ne... | Update summarization is the problem of extracting and synthesizing novel information in a collection of documents with respect to a set of documents assumed to be known by the reader. This problem has received much attention in recent years, as can be observed in the number of participants to the special track on updat... | Most Bayesian approaches to summarization are based on topic models. These generative models represent documents as mixtures of latent topics, where a topic is a probability distribution over words. In TOPICSUM A commonality of all these models is the use of collection and document-specific distributions in order to di... | 693 | 1,375 | 693 |
How to Ask Good Questions? Try to Leverage Paraphrases | Given a sentence and its relevant answer, how to ask good questions is a challenging task, which has many real applications. Inspired by human's paraphrasing capability to ask questions of the same meaning but with diverse expressions, we propose to incorporate paraphrase knowledge into question generation(QG) to gener... | Question generation (QG) is an essential task for NLP, which focuses on generating grammatical questions for given paragraphs or sentences. It plays a vital role in various realistic scenarios. For educational purposes, QG can create reading comprehension materials for language learners tems Recent neural network-based... | For current mainstream neural network-based methods on QG, most approaches utilize the Seq2Seq model with attention mechanism In order to make use of the context information of paragraphs, Paraphrase knowledge has been used to improve many NLP tasks, such as machine translation, ques-tion answering, and text simplifica... | 1,049 | 1,139 | 1,049 |
Generating Summaries with Topic Templates and Structured Convolutional Decoders | Existing neural generation approaches create multi-sentence text as a single sequence. In this paper we propose a structured convolutional decoder that is guided by the content structure of target summaries. We compare our model with existing sequential decoders on three data sets representing different domains. Automa... | Abstractive multi-document summarization aims at generating a coherent summary from a cluster of thematically related documents. Recently, Like most previous work on neural text generation In this work we propose a neural model which is guided by the topic structure of target summaries, i.e., the way content is organiz... | The Wikipedia lead section introduces the entity (e.g., Country or Brazil) the article is about, highlighting important facts associated with it. We explicitly model the topic structure of summaries, under the assumption that documents cover different topics about a given entity, while the summary covers the most salie... | 405 | 2,424 | 405 |
How effective is BERT without word ordering? Implications for language understanding and data privacy | Ordered word sequences contain the rich structures that define language. However, it's often not clear if or how modern pretrained language models utilize these structures. We show that the token representations and selfattention activations within BERT are surprisingly resilient to shuffling the order of input tokens,... | Masked language models (MLMs) like BERT To assess this question, we first compare the internal representations of BERT and RoBERTa Following cues from prior work The bad news: Despite BERT being trained on intact word sequences, BoW-BERT demonstrates that MLMs can readily ignore syntax (while maintaining strong perform... | Shuffling inputs to non-pretrained models. Word order shuffling has been tested as part of the full training process for non-pretrained models. Shuffling inputs to pretrained MLMs. While at the time of submission of this work, shuffling results had not been fully reported on the popular GLUE taskset, prior results have... | 809 | 571 | 809 |
Exploring Cross-lingual Textual Style Transfer with Large Multilingual Language Models | Detoxification is a task of generating text in polite style while preserving meaning and fluency of the original toxic text. Existing detoxification methods are designed to work in one exact language. This work investigates multilingual and cross-lingual detoxification and the behavior of large multilingual models like... | The task of Textual Style Transfer (Textual Style Transfer) can be viewed as a task where certain properties of text are being modified while rest retain the same Some examples of detoxification presented in Table Textual style transfer gained a lot of attention with a rise of deep learning-based NLP methods. Given tha... | Target text What the f*ck is your problem? What is your problem? This whole article is bullshit. This article is not good. Yeah, this clowns gonna make alberta great again! Yeah, this gonna make Alberta great again Multilingual language models such as mBART Our contributions can be summarized as follows 1. We introduce... | 695 | 959 | 695 |
Self-Improvement of Non-autoregressive Model via Sequence-Level Distillation | Although Non-autoregressive Transformer (NAT) models have achieved great success in terms of fast inference speed, this speedup comes with a performance drop due to the inherent multi-modality problem of the NAT model. Previous works commonly alleviate this problem by replacing the target side of the raw data with dist... | Non-autoregressive Transformer (NAT) models One way to improve the ability of NAT models to handle complex data is by enhancing their capacity Another common approach to alleviate the multi-modality problem is modifying the target sequence Therefore, most existing works only consider knowledge distillation as a necessa... | In this section, we first briefly describe the task formulation and then introduce three types of NAT models. The machine translation task can be formally defined as a sequence-to-sequence generation problem. Given the target language sequence y={y 1 , y 2 , ..., y T } and source language sequence x={x 1 , x 2 , ..., x... | 1,192 | 1,365 | 1,192 |
Relational Word Embeddings | While word embeddings have been shown to implicitly encode various forms of attributional knowledge, the extent to which they capture relational information is far more limited. In previous work, this limitation has been addressed by incorporating relational knowledge from external knowledge bases when learning the wor... | Word embeddings are paramount to the success of current natural language processing (NLP) methods. Apart from the fact that they provide a convenient mechanism for encoding textual information in neural network models, their importance mainly stems from the remarkable amount of linguistic and semantic information that ... | Relation Vectors. A number of approaches have been proposed that are aimed at learning relation vectors for a given set of word pairs (a,b), based on sentences in which these word pairs co-occur. For instance, Turney (2005) introduced a method called Latent Relational Analysis (LRA), which relies on first identifying a... | 968 | 2,237 | 968 |
Learning-Based Named Entity Recognition for Morphologically-Rich, Resource-Scarce Languages | Named entity recognition for morphologically rich, case-insensitive languages, including the majority of semitic languages, Iranian languages, and Indian languages, is inherently more difficult than its English counterpart. Worse still, progress on machine learning approaches to named entity recognition for many of the... | While research in natural language processing has gained a lot of momentum in the past several decades, much of this research effort has been focusing on only a handful of politically-important languages such as English, Chinese, and Arabic. On the other hand, being the fifth most spoken language One potential solution... | Cucerzan and Yarowsky (1999) exploit morphological and contextual patterns to propose a language-independent solution to NER. They use affixes based on the paradigm that named entities corresponding to a particular class have similar morphological structure. Their bootstrapping approach is tested on Romanian, English, ... | 1,192 | 3,894 | 1,192 |
Local Structure Matters Most in Most Languages | Many recent perturbation studies have found unintuitive results on what does and does not matter when performing Natural Language Understanding (NLU) tasks in English. Coding properties, such as the order of words, can often be removed through shuffling without impacting downstream performances. Such insight may be use... | A recent research trend has explored the sensitivity, or insensitivity, of neural language models to different perturbations of texts One such coding property, the local structure of text, has recently been shown to be ubiquitously relied upon by both neural language models In this short paper, our contributions are as... | We extend the perturbation studies of The CHRF-2 (chrF) The Index Displacement Count (IDC) The compression rate (Comp) The scholar is typesetting. We perform perturbations by altering the order of subwords and characters present in the text. Three types of perturbations are applied. Full shuffling completely randomizes... | 838 | 589 | 838 |
Arabic Morphology Generation Using a Concatenative Strategy | Arabic inflectional morphology requires infixation, prefixation and suffixation, giving rise to a large space of morphological variation. In this paper we describe an approach to reducing the complexity of Arabic morphology generation using discrimination trees and transformational rules. By decoupling the problem of s... | Morphologically, Arabic is a non-concatenative language. The basic problem with generating Arabic verbal morphology is the large number of variants that must be generated. Verbal stems are based on triliteral or quadriliteral roots (3-or 4-radicals). Stems are formed by a derivational combination of a root morpheme and... | Verb roots in Arabic can be classified as shown in Figure Strong verbs undergo systematic changes in stem voweling from the perfect to the imperfect. The first radical vowel disappears in the imperfect. Verbs whose middle radical vowel in the perfect is 'a' can change it to 'a' (e.g., qaTa'a 'he cut' -> yaqTa'u 'he cut... | 563 | 3,419 | 563 |
MPC-BERT: A Pre-Trained Language Model for Multi-Party Conversation Understanding | Recently, various neural models for multiparty conversation (MPC) have achieved impressive improvements on a variety of tasks such as addressee recognition, speaker identification and response prediction. However, these existing methods on MPC usually represent interlocutors and utterances individually and ignore the i... | Building a conversational agent with intelligence has drawn significant attention from both academia and industry. Most of existing methods have studied understanding conversations between two participants, aiming to return an appropriate response either in a generation-based 2015; An instance of MPC always contains co... | Existing methods on building dialogue systems can be generally categorized into studying twoparty conversations and multi-party conversations (MPC). In this paper, we study MPC. In addition to predicting utterances, identifying the speaker and recognizing the addressee of an utterance are also important tasks for MPC. ... | 1,286 | 3,061 | 1,286 |
Multiple Tasks Integration: Tagging, Syntactic and Semantic Parsing as a Single Task | Departing from both sequential pipelines and monotask systems, we propose Multiple Tasks Integration (MTI), a multitask paradigm orthogonal to weight sharing. The essence of MTI is to process the input iteratively but concurrently at multiple levels of analysis, where each decision is based on all of the structures tha... | Historically, Natural Language Processing (NLP) systems have generally been built as sequential pipelines, where each module adds another layer of annotation, in order of (supposed) increasing complexity. Progress in neural networks has, however, led to the development of state-of-the-art systems that completely bypass... | Fighting error propagation Since its apparition for speech recognition Other techniques are designed to help to fight error propagation of the second kind, which focus on the training of the system to make it robust to its own mistakes. One possibility consists in training the system to predict the next best action in ... | 882 | 2,452 | 882 |
Using Second-order Vectors in a Knowledge-based Method for Acronym Disambiguation | In this paper, we introduce a knowledge-based method to disambiguate biomedical acronyms using second-order co-occurrence vectors. We create these vectors using information about a long-form obtained from the Unified Medical Language System and Medline. We evaluate this method on a dataset of 18 acronyms found in biome... | W ord Sense Disambiguation (WSD) is the task of automatically identifying the appropriate sense of a word with multiple senses. For example, the word culture could refer to anthropological culture (e.g., the culture of the Mayan civilization), or a laboratory culture (e.g., cell culture). Acronym disambiguation is the ... | The Unified Medical Language System (UMLS) is a data warehouse that stores a number of distinct biomedical and clinical resources. One such resource, used in this work, is the Metathesaurus. The Metathesaurus contains biomedical and clinical concepts from over 100 disparate terminology sources that have been semi-autom... | 531 | 2,095 | 531 |
OssCSE: Overcoming Surface Structure Bias in Contrastive Learning for Unsupervised Sentence Embedding | Contrastive learning has been demonstrated effective in unsupervised sentence representation learning. Given one sentence, positive pairs are obtained by passing the sentence to the encoder twice using the different dropout masks, and negative pairs are obtained by taking another sentence in the same mini-batch. Howeve... | Deep and surface structures Recent studies To answer the first question, we first propose to validate whether current models can correctly rank a few sentence transformations. Furthermore, to systematically evaluate how the bias effect existing models, we split the existing datasets following the Consistency (Cont.) an... | Unsupervised sentence representation learning has been widely studied. The early study attempts to leverage sentence internal structure Recently, unsupervised sentence embeddings have utilized contrastive learning schemes to further boost the performance by different data augmentation methods, such as dropout To speci... | 1,552 | 1,751 | 1,552 |
STRUDEL : Structured Dialogue Summarization for Dialogue Comprehension | Abstractive dialogue summarization has long been viewed as an important standalone task in natural language processing, but no previous work has explored the possibility of whether abstractive dialogue summarization can also be used as a means to boost an NLP system's performance on other important dialogue comprehensi... | In natural language processing, abstractive dialogue summarization In this paper, we propose a novel type of dialogue summarization task -STRUctured DiaLoguE Summarization (STRUDEL 1 ) -that can help pre-trained language models to better understand dialogues and improve their performance on important dialogue comprehen... | Abstractive summarization aims to generate a concise summary of a text by producing a paraphrasing of the main contents using different vocabulary, rather than simply extracting the important sentences, which is referred to as extractive summarization. A popular approach to produce abstractive summaries of long documen... | 1,555 | 1,135 | 1,555 |
Sampling-Based Approximations to Minimum Bayes Risk Decoding for Neural Machine Translation | In NMT we search for the mode of the model distribution to form predictions. The mode and other high-probability translations found by beam search have been shown to often be inadequate in a number of ways. This prevents improving translation quality through better search, as these idiosyncratic translations end up sel... | NMT systems Eikema and In this work, we first analyse the procedure by Eikema and | NMT employs neural networks (NNs) to predict a conditional probability distribution Y |θ, x over translation candidates of any given source sentence x. The sample space Y is the set of all sequences of known target-language symbols (e.g., sub-word units). NMT factorises the distribution as a chain of random draws from ... | 1,175 | 81 | 1,175 |
Candidate Soups: Fusing Candidate Results Improves Translation Quality for Non-Autoregressive Translation | Non-autoregressive translation (NAT) model achieves a much faster inference speed than the autoregressive translation (AT) model because it can simultaneously predict all tokens during inference. However, its translation quality suffers from degradation compared to AT. And existing NAT methods only focus on improving t... | Autoregressive translation (AT) models based on Transformer Therefore, the non-autoregressive translation (NAT) Several methods have been proposed to alleviate the multimodality problem and improve the performance of the NAT model, such as the iterationbased NAT model Most of the previous methods are modified from the ... | It often costs over a hundred dollars to obtain the required identity card . Candidate Soups, which can significantly improve the translation quality without any modification to the model. Moreover, Candidate Soups is a general approach that can be used by any NAT model that can generate multiple candidate results, suc... | 1,062 | 384 | 1,062 |
Lifelong Sequence Generation with Dynamic Module Expansion and Adaptation | Lifelong sequence generation (LSG), a problem in continual learning, aims to continually train a model on a sequence of generation tasks to learn constantly emerging new generation patterns while avoiding the forgetting of previous knowledge. Existing LSG methods mainly focus on maintaining old knowledge while paying l... | With the recent advancements in pre-trained language models (LMs), current sequence generation methods have achieved impressive performance on a variety of generation tasks A potential solution is to formalize sequence generation as lifelong sequence generation or LSG Despite its effectiveness, ACM has several key limi... | Lifelong Learning (LL) aims to continually learn knowledge from a sequence of tasks with different distributions. The goal is twofold: alleviate catastrophic forgetting Catastrophic forgetting typically means that the model forgets previously acquired knowledge after learning new tasks. Prior LL methods mainly focus on... | 1,170 | 1,233 | 1,170 |
Learning the Beauty in Songs: Neural Singing Voice Beautifier | We are interested in a novel task, singing voice beautification (SVB). Given the singing voice of an amateur singer, SVB aims to improve the intonation and vocal tone of the voice, while keeping the content and vocal timbre. Current automatic pitch correction techniques are immature, and most of them are restricted to ... | The major successes of the artificial intelligent singing voice research are primarily in Singing Voice Synthesis (SVS) Nowadays in real-life scenarios, SVB is usually performed by professional sound engineers with adequate domain knowledge, who manipulate commercial vocal correction tools such as Melodyne To tackle th... | Singing Voice Conversion (SVC) is a sub-task of Voice Conversion (VC) Mainstream SVC models can be grouped into three categories Automatic Pitch Correction (APC) works attempt to minimize the manual effort in modifying the flawed singing voice In this section, we describe the overview of NSVB, which is shown in Figure ... | 602 | 1,922 | 602 |
Why Should Adversarial Perturbations be Imperceptible? Rethink the Research Paradigm in Adversarial NLP WARNING: This paper contains real-world cases which are offensive in nature | Textual adversarial samples play important roles in multiple subfields of NLP research, including security, evaluation, explainability, and data augmentation. However, most work mixes all these roles, obscuring the problem definitions and research goals of the security role that aims to reveal the practical concerns of... | Natural language processing (NLP) models based on deep learning have been employed in many realworld applications | Adversarial samples can reveal the practical concerns of NLP models deployed in security situations. Adversarial samples can be employed to benchmark models' robustness to out-of-distribution data (diverse user inputs). Adversarial samples can explain part of the models' decision processes. Adversarial training based o... | 1,364 | 113 | 1,364 |
Improving Tokenisation by Alternative Treatment of Spaces | Tokenisation is the first step in almost all NLP tasks, and state-of-the-art transformer-based language models all use subword tokenisation algorithms to process input text. Existing algorithms have problems, often producing tokenisations of limited linguistic validity and representing equivalent strings differently de... | Tokenisation is a key initial step in processing natural language, as it identifies the linguistic units to be processed, converting them to numerical IDs which can then be vectorised and manipulated by mathematical operations. Earlier NLP approaches used simple stringsearching techniques with regular expressions to to... | Existing tokenisation algorithms often produce unintuitive tokenisations for complex words, incorrectly splitting prefixes, and producing unmeaningful subword tokens, which are problems that have been discussed in previous works. For these latter examples, there is a second problem: even if the base were tokenised as a... | 1,241 | 2,138 | 1,241 |
Demonstration of a Neural Machine Translation System with Online Learning for Translators | We introduce a demonstration of our system, which implements online learning for neural machine translation in a production environment. These techniques allow the system to continuously learn from the corrections provided by the translators. We implemented an end-to-end platform integrating our machine translation ser... | Productivity is crucial in the translation industry. Nowadays, translation companies must be more competitive than ever and meet fast commercial demands. Thus, they need to produce high quality translations in shorter periods of time. Machine translation (MT) can help them to achieve this goal: instead of a linguist th... | We are interested in benefiting from the post-edits generated by the user during the PE process. To that end, we update the system on-the-fly, i.e, as soon as a sentence has been validated by the posteditor. Right after the user confirms a post-edit, we update the models of our NMT system, using the source sentence and... | 580 | 1,771 | 580 |
Dual Slot Selector via Local Reliability Verification for Dialogue State Tracking | The goal of dialogue state tracking (DST) is to predict the current dialogue state given all previous dialogue contexts. Existing approaches generally predict the dialogue state at every turn from scratch. However, the overwhelming majority of the slots in each turn should simply inherit the slot values from the previo... | Task-oriented dialogue has attracted increasing attention in both the research and industry communities. As a key component in task-oriented dialogue systems, Dialogue State Tracking (DST) aims to extract user goals or intents and represent them as a compact dialogue state in the form of slot-value pairs of each turn d... | Traditional statistical dialogue state tracking models combine semantics extracted by spoken language understanding modules to predict the current dialogue state On the other hand, dialogue state tracking and machine reading comprehension (MRC) have similarities in many aspects Figure Dual Slot Selector the dialogue st... | 1,465 | 2,857 | 1,465 |
Opportunistic Decoding with Timely Correction for Simultaneous Translation | Simultaneous translation has many important application scenarios and attracts much attention from both academia and industry recently. Most existing frameworks, however, have difficulties in balancing between the translation quality and latency, i.e., the decoding policy is usually either too aggressive or too conserv... | Simultaneous translation, which starts translation before the speaker finishes, is extremely useful in many scenarios, such as international conferences, travels, and so on. In order to achieve low latency, it is often inevitable to generate target words with insufficient source information, which makes this task extre... | Full-sentence NMT. The conventional fullsentence NMT processes the source sentence x = (x 1 , ..., x n ) with an encoder, where x i represents an input token. The decoder on the target side (greedily) selects the highest-scoring word y t given source representation h and previously generated target tokens, y <t = (y 1 ... | 905 | 2,101 | 905 |
Using sparse semantic embeddings learned from multimodal text and image data to model human conceptual knowledge | Distributional models provide a convenient way to model semantics using dense embedding spaces derived from unsupervised learning algorithms. However, the dimensions of dense embedding spaces are not designed to resemble human semantic knowledge. Moreover, embeddings are often built from a single source of information ... | Distributional Semantic Models (DSMs) are used to represent semantic information about concepts in a high-dimensional vector space, where each concept is represented as a point in the space such that concepts with more similar meanings are closer together. Unsupervised learning algorithms are regularly employed to prod... | Much of the research aimed at the sparse decomposition of dense vector spaces is closely associated with the work of In total, we used sixteen distributional semantic models, eight of which are dense and eight of which are their sparse counterparts. These models are summarized in Table We implemented two state-of-the-a... | 911 | 971 | 911 |
Scene Graph as Pivoting: Inference-time Image-free Unsupervised Multimodal Machine Translation with Visual Scene Hallucination | In this work, we investigate a more realistic unsupervised multimodal machine translation (UMMT) setup, inference-time image-free UMMT, where the model is trained with sourcetext image pairs, and tested with only sourcetext inputs. First, we represent the input images and texts with the visual and language scene graphs... | Current neural machine translation (NMT) has achieved great triumph UMMT systems are trained with only the textimage pairs (<text-img>), which can be easier to collect than the parallel source-target sentence pairs (<src-tgt>) In this work, we present a novel UMMT method that solves all aforementioned challenges. First... | Neural machine translation has achieved notable development in the era of deep learning Unsupervised machine translation aims to learn cross-lingual mapping without the use of largescale parallel corpora. The setting is practically meaningful to those minor languages with hard data accessibility. The basic idea is to l... | 1,017 | 908 | 1,017 |
Improving Topic Quality by Promoting Named Entities in Topic Modeling | News-related content has been extensively studied in both topic modeling research and named entity recognition. However, expressive power of named entities and their potential for improving the quality of discovered topics has not received much attention. In this paper we use named entities as domain-specific terms for... | News-centric content conveys information about events, individuals and other entities. Analysis of news-related documents includes identifying hidden features for classifying them or summarizing the content. Topic modeling is the standard technique for such purposes, and Latent Dirichlet Allocation (LDA) The main contr... | This section describes the related work in the area of topic modeling, specifically LDA. Several works explored the relation between LDA and named entities in recent years. The most famous model is CorrLDA2 Traditionally, the input of LDA is a documentterm matrix of term frequencies (TF), according to the bag-of-words ... | 611 | 1,312 | 611 |
A General-Purpose Algorithm for Constrained Sequential Inference | Inference in structured prediction involves finding the best output structure for an input, subject to certain constraints. Many current approaches use sequential inference, which constructs the output in a left-to-right manner. However, there is no general framework to specify constraints in these approaches. We prese... | The key challenge in structured prediction problems (like sequence tagging and parsing) is inference (also known as decoding), which involves identifying the best output structure y for an input instance x from an exponentially large search space Y At present, inference algorithms are designed to handle task-specific c... | We briefly review automata that we use for representing constraints in our algorithm. For the purposes of this work, an automaton is a (possibly weighted) directed graph that compactly encodes a set of strings, known as its language. The two types of automata used in this work are finite-state automata (FSA) and push-d... | 1,318 | 2,867 | 1,318 |
Event Embeddings for Semantic Script Modeling | Semantic scripts is a conceptual representation which defines how events are organized into higher level activities. Practically all the previous approaches to inducing script knowledge from text relied on count-based techniques (e.g., generative models) and have not attempted to compositionally model events. In this w... | It is generally believed that the lack of knowledge on how individual events are organized into higher-level scenarios is one of the major obstacles for natural language understanding. Texts often do not provide a detailed specification of underlying events as writers rely on the ability of humans to read between the l... | The general idea in the previous count based methods is to collect events sequences for an entity from the corpus (referred as a script). An entity is typically a noun/pronoun describing a person, location or temporal construct mentioned in a document. A document is parsed using a statistical dependency parser. Then, t... | 686 | 2,079 | 686 |
Don't Parse, Choose Spans! Continuous and Discontinuous Constituency Parsing via Autoregressive Span Selection | We present a simple and unified approach for both continuous and discontinuous constituency parsing via autoregressive span selection. Constituency parsing aims to produce a set of non-crossing spans so that they can form a constituency parse tree. We sort gold spans in a predefined order and train a pointer network to... | Constituency parsing is a fundamental task in natural language processing, having many applications in downstream tasks such as language modeling Both continuous and discontinuous parsing can be framed as span prediction problems. In continuous parsing, each span corresponds to a single interval (of the observed senten... | We formally frame (dis)continuous constituency parsing as a span selection problem. A (dis)continuous constituency parse tree t comprises a set of nodes and for each node s we have yield(s) = {s 1 , ..., s l } which is the set of sorted token indices in the yield of s in t with s 1 < • • • < s l . The bidirectional con... | 1,037 | 1,372 | 1,037 |
Dialogue Natural Language Inference | Consistency is a long standing issue faced by dialogue models. In this paper, we frame the consistency of dialogue agents as natural language inference (NLI) and create a new natural language inference dataset called Dialogue NLI. We propose a method which demonstrates that a model trained on Dialogue NLI can be used t... | A long standing issue faced by dialogue models is consistency One approach to increasing the consistency of a chit-chat dialogue model was proposed in Separately, the framework of Natural Language Inference (NLI) Despite this expectation, leveraging an NLI model for a downstream task remains an underexplored research d... | Language Inference First, we review the dialogue generation and natural language inference problems as well as the notions of consistency used throughout. Dialogue Generation Dialogue generation can be framed as next utterance prediction, in which an utterance (a sequence of tokens representing a sentence) u t+1 is pr... | 521 | 1,151 | 521 |
Beyond the Granularity: Multi-Perspective Dialogue Collaborative Selection for Dialogue State Tracking | In dialogue state tracking, dialogue history is a crucial material, and its utilization varies between different models. However, no matter how the dialogue history is used, each existing model uses its own consistent dialogue history during the entire state tracking process, regardless of which slot is updated. Appare... | Task-oriented dialogue systems have recently attracted growing attention and achieved substantial progress. Dialogue state tracking (DST) is a core component, where it is responsible for interpreting user goals and intents and feeding hotel-type: | ... Figure In dialogue state tracking, dialogue history is a crucial source material. Recently, granularity has been proposed to quantify the utilization of dialogue history Furtherly, granularity means directly working on all dialogue contents from a particular turn to the current turn, regardless of the fact that the... | 1,466 | 246 | 1,466 |
Uncovering divergent linguistic information in word embeddings with lessons for intrinsic and extrinsic evaluation | Following the recent success of word embeddings, it has been argued that there is no such thing as an ideal representation for words, as different models tend to capture divergent and often mutually incompatible aspects like semantics/syntax and similarity/relatedness. In this paper, we show that each embedding model c... | Word embeddings have recently become a central topic in natural language processing. Several unsupervised methods have been proposed to efficiently train dense vector representations of words While there is still an active research line to better understand these models from a theoretical perspective Nevertheless, the ... | Let X be the matrix of word embeddings in a given language, so that X i * is the embedding of the ith word in the vocabulary. Such embeddings are meant to capture the meaning of their corresponding words in such a way that the dot product sim(i, j) = X i * • X j * gives some measure of the similarity between the ith an... | 813 | 2,401 | 813 |
The GATE Crowdsourcing Plugin: Crowdsourcing Annotated Corpora Made Easy | Crowdsourcing is an increasingly popular, collaborative approach for acquiring annotated corpora. Despite this, reuse of corpus conversion tools and user interfaces between projects is still problematic, since these are not generally made available. This demonstration will introduce the new, open-source GATE Crowdsourc... | Annotation science A big outstanding challenge for crowdsourcing projects is that the cost to define a single annotation task remains quite substantial. This demonstration will introduce the new, open-source GATE Crowdsourcing plugin, which offers infrastructural support for mapping documents to crowdsourcing units, as... | Conceptually, the process of crowdsourcing annotated corpora can be broken down into four main stages, within which there are a number of largely infrastructural steps. In particular, data preparation and transformation into CrowdFlower units, creation of the annotation UI, creation and upload of gold units for quality... | 737 | 378 | 737 |
Latent Variable Model for Multi-modal Translation | In this work, we propose to model the interaction between visual and textual features for multi-modal neural machine translation (MMT) through a latent variable model. This latent variable can be seen as a multi-modal stochastic embedding of an image and its description in a foreign language. It is used in a target-lan... | Multi-modal machine translation (MMT) is an exciting novel take on machine translation (MT) where we are interested in learning to translate sentences in the presence of visual input (mostly images). In the last three years there have been shared tasks Most MMT models expand neural machine translation (NMT) architectur... | Similarly to standard NMT, in MMT we wish to translate a source sequence The main difference is the presence of an image v which illustrates the sentence pair x m 1 , y n 1 . We do not model images directly, but instead an 2048dimensional vector of pre-activations of a ResNet-50's pool5 layer In our variational MMT mod... | 728 | 1,727 | 728 |
Overlap-based Vocabulary Generation Improves Cross-lingual Transfer Among Related Languages | Pre-trained multilingual language models such as mBERT and XLM-R have demonstrated great potential for zero-shot cross-lingual transfer to low web-resource languages (LRL). However, due to limited model capacity, the large difference in the sizes of available monolingual corpora between high web-resource languages (HRL... | Zero-shot cross-lingual transfer is the ability of a model to learn from labeled data in one language and transfer the learning to another language without any labeled data. Transformer Vocabulary generation is an important step in multilingual model training, where vocabulary size directly impacts model capacity. Usua... | Transformer-based multilingual language models such as mBERT Input Data In the data creation stage, Tokenization Vocabulary Generation We are not aware of any prior work that explicitly promotes overlapping tokens between LRLs and HRLs in the vocabulary of multilingual models. We are given monolingual data D 1 , ..., D... | 1,371 | 2,929 | 1,371 |
Deriving Generalized Knowledge from Corpora using WordNet Abstraction | Existing work in the extraction of commonsense knowledge from text has been primarily restricted to factoids that serve as statements about what may possibly obtain in the world. We present an approach to deriving stronger, more general claims by abstracting over large sets of factoids. Our goal is to coalesce the obse... | Our interest is ultimately in building systems with commonsense reasoning and language understanding abilities. As is widely appreciated, such systems will require large amounts of general world knowledge. Large text corpora are an attractive potential source of such knowledge. However, current natural language underst... | Rilly or Glendora had entered her room while she slept, bringing back her washed clothes. (: Here the upper-case sentences are automatically generated verbalizations of the abstracted LFs shown beneath them. The goal in this work, with respect to the example given, would be to derive with the use of a large collection ... | 558 | 1,989 | 558 |
What to Read in a Contract? Party-Specific Summarization of Legal Obligations, Entitlements, and Prohibitions | Reviewing and comprehending key obligations, entitlements, and prohibitions in legal contracts can be a tedious task due to their length and domain-specificity. Furthermore, the key rights and duties requiring review vary for each contracting party. In this work, we propose a new task of party-specific extractive summa... | A contract is a legally binding agreement that defines and governs the rights, duties, and responsibilities of all parties involved in it. To sign a contract (e.g., lease agreements, terms of services, and privacy policies), it is important for these parties to precisely understand their rights and duties as described ... | Summarization of Legal Text Existing works focus on summarizing legal case reports Existing works either propose rule-based methods We formally define the new task as: given a contract C consisting of a sequence of sentences (c 1 , c 2 , . . . , c n ) and a party P , the task is to generate an extractive summary S cons... | 1,048 | 2,702 | 1,048 |
Flexible Visual Grounding | Existing visual grounding datasets are artificially made, where every query regarding an entity must be able to be grounded to a corresponding image region, i.e., answerable. However, in real-world multimedia data such as news articles and social media, many entities in the text cannot be grounded to the image, i.e., u... | Starting from conventional vision-and-language tasks such as image captioning VQA, it is crucial to understand to which image region the question is referring. Because of the importance of visual grounding, many research efforts have been dedicated to improve its accuracy Previous visual grounding work assume that a qu... | Previous visual grounding studies have been conducted on different datasets. In the Flickr30k entities dataset Regarding visual grounding models, Inspired by the success of pre-training language models such as BERT Because there are no existing visual grounding datasets where unanswerable queries are contained, we pres... | 543 | 2,651 | 543 |
Mixed-Lingual Pre-training for Cross-lingual Summarization | Cross-lingual Summarization (CLS) aims at producing a summary in the target language for an article in the source language. Traditional solutions employ a twostep approach, i.e. translate→summarize or summarize→translate. Recently, end-to-end models have achieved better results, but these approaches are mostly limited ... | Text summarization can facilitate the propagation of information by providing an abridged version for long articles and documents. Meanwhile, the globalization progress has prompted a high demand of information dissemination across language barriers. Thus, the cross-lingual summarization (CLS) task emerges to provide a... | Pre-training language models Early literatures on cross-lingual summarization focus on the two-step approach involving machine translation and summarization (2018) presents a solution to zero-shot cross-lingual headline generation by using machine translation and summarization datasets. We propose a set of multi-task ... | 1,085 | 1,041 | 1,085 |
Equipping Language Models with Tool Use Capability for Tabular Data Analysis in Finance | Large language models (LLMs) have exhibited an array of reasoning capabilities but face challenges like error propagation and hallucination, particularly in specialised areas like finance, where data is heterogeneous, and precision is paramount. We explore the potential of language model augmentation with external tool... | Augmenting Large Language Models (LLMs) with tools has emerged as a promising approach to further complement LLMs' capabilities with specialised mechanisms, leading to improved accuracy and reliability This paradigm holds particular appeal in fields demanding precision, such as finance A satisfying review of existing w... | We use the LLAMA 2 13B CHAT We use a mixture of four financial and generic structured and unstructured question-answering datasets. We provide a brief summary in below. TAT-QA. Consists of questions generated by financial experts associated with hybrid contexts drawn from real-world financial reports Financial PhraseBa... | 1,146 | 1,527 | 1,146 |
Accelerating Neural Transformer via an Average Attention Network | With parallelizable attention networks, the neural Transformer is very fast to train. However, due to the auto-regressive architecture and self-attention in the decoder, the decoding procedure becomes slow. To alleviate this issue, we propose an average attention network as an alternative to the self-attention network ... | The past few years have witnessed the rapid development of neural machine translation (NMT), which translates a source sentence into the target language with an encoder-attention-decoder framework Most interestingly, the neural Transformer is capable of being fully parallelized at the training phase and modeling intra-... | GRU The attention mechanism is originally proposed to induce translation-relevant source information for predicting next target word in NMT. It contributes a lot to make NMT outperform SMT. Recently, a variety of efforts are made to further improve its accuracy and capability. In the respect of speeding up the decoding... | 1,111 | 2,278 | 1,111 |
Generating Long and Informative Reviews with Aspect-Aware Coarse-to-Fine Decoding | Generating long and informative review text is a challenging natural language generation task. Previous work focuses on word-level generation, neglecting the importance of topical and syntactic characteristics from natural languages. In this paper, we propose a novel review generation model by characterizing an elabora... | In the past decades, online review services (e.g., In the literature, various methods have been developed for review generation As found in the literature of linguistics Based on such a generation process, in this paper, we propose a novel aspect-aware coarse-tofine decoder for generating product reviews. We first util... | In recent years, researchers have made great progress in natural language generation (NLG) It has been found that RNN models tend to generate short, repetitive, and dull texts Our work is inspired by the work of using sketches as intermediate representations A review is a natural language text written by a user u on a ... | 834 | 1,592 | 834 |
Does Your Model Classify Entities Reasonably? Diagnosing and Mitigating Spurious Correlations in Entity Typing | Entity typing aims at predicting one or more words that describe the type(s) of a specific mention in a sentence. Due to shortcuts from surface patterns to annotated entity labels and biased training, existing entity typing models are subject to the problem of spurious correlations. To comprehensively investigate the f... | Given a sentence with an entity mention, the entity typing task aims at predicting one or more words or phrases that describe the type(s) of that specific mention To tackle the task, literature has developed vari-ous predictive methods to capture the association between the contextualized entity mention representation ... | In this section, we start with the problem definition ( §2.1) and then categorize and diagnose the spurious correlations causing shortcut predictions by the typing model ( §2.2). Lastly, we propose a counterfactual data augmentation approach to mitigate the identified spurious correlations, as well as several alternati... | 1,250 | 907 | 1,250 |
Reasoning with Latent Structure Refinement for Document-Level Relation Extraction | Document-level relation extraction requires integrating information within and across multiple sentences of a document and capturing complex interactions between inter-sentence entities. However, effective aggregation of relevant information in the document remains a challenging research question. Existing approaches c... | Relation extraction aims to detect relations among entities in the text and plays a significant role in a variety of natural language processing applications. Early research efforts focus on predicting relations between entities within the sentence A more challenging, yet practical extension, is the document-level rela... | Node constructor encodes sentences in a document into contextual representations and constructs representations of mention nodes, entity nodes and meta dependency paths (MDP) nodes, as shown in Figure Given a document d, each sentence d i in it is fed to the context encoder, which outputs the contextualized representat... | 1,183 | 1,741 | 1,183 |
Metaphors in Pre-Trained Language Models: Probing and Generalization Across Datasets and Languages | Human languages are full of metaphorical expressions. Metaphors help people understand the world by connecting new concepts and domains to more familiar ones. Large pretrained language models (PLMs) are therefore assumed to encode metaphorical knowledge useful for NLP systems. In this paper, we investigate this hypothe... | Pre-trained language models (PLMs) Metaphors are important aspects of human languages. In conceptual metaphor theory (CMT) So far, there has been no comprehensive analysis of whether and how PLMs represent metaphorical information. We intuitively assume that PLMs must encode some information about metaphors due to thei... | Metaphor detection using PLMs. The metaphor detection task Probing methods in NLP. Probing is an analytical tool used for assessing linguistic knowledge in language representations. In probing, the information richness of the representations is inspected by the quality of a supervised model in predicting linguistic pro... | 963 | 1,843 | 963 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.