Iterative data generation and model re-training can effectively align large language models (LLMs) to human preferences. The process of data sampling is crucial, as it significantly influences the success of policy improvement. Repeated random sampling is a widely used method that independently queries the model multiple times to generate outputs. In this work, we propose a more effective sampling method, named Preference-Guided Reflective Sampling (PRS). Unlike random sampling, PRS employs a tree-based generation framework to enable more efficient sampling. It leverages adaptive self-refinement techniques to better explore the sampling space. By specifying user preferences in natural language, PRS can further optimize response generation according to these preferences. As a result, PRS can align models to diverse user preferences. Our experiments demonstrate that PRS generates higher-quality responses with significantly higher rewards. On AlpacaEval and Arena-Hard, PRS substantially outperforms repeated random sampling in best-of-N sampling. Moreover, PRS shows strong performance when applied in iterative offline RL training.

Knowledge in the real world is being updated constantly. However, it is costly to frequently update large language models (LLMs). Therefore, it is crucial for LLMs to understand the concept of temporal knowledge. However, prior works on temporal question answering (TQA) did not emphasize multi-answer and multi-hop types of temporal reasoning. In this paper, we propose a complex temporal question-answering dataset Complex-TR that focuses on multi-answer and multi-hop temporal reasoning. Besides, we also propose a novel data augmentation strategy to improve the complex temporal reasoning capability and robustness of LLMs. We conducted experiments on multiple temporal QA datasets. Experimental results show that our method is able to improve LLMs’ performance on temporal QA benchmarks by significant margins.

The natural language processing field has been evolving around language models for the past few years, from the usage of n-gram language models for re-ranking, to transfer learning with encoder-only (BERT-like) language models, and finally to large language models (LLMs) as general solvers. LLMs are dominated by the decoder-only type, and they are popular for their efficacy in numerous tasks. LLMs are regarded as having strong comprehension abilities and strong capabilities to solve new unseen tasks. As such, people may quickly assume that decoder-only LLMs always perform better than the encoder-only ones, especially for understanding word meaning. In this paper, we demonstrate that decoder-only LLMs perform worse on word meaning comprehension than an encoder-only language model that has vastly fewer parameters.

Error type information has been widely used to improve the performance of grammatical error correction (GEC) models, whether for generating corrections, re-ranking them, or combining GEC models. Combining GEC models that have complementary strengths in correcting different error types is very effective in producing better corrections. However, system combination incurs a high computational cost due to the need to run inference on the base systems before running the combination method itself. Therefore, it would be more efficient to have a single model with multiple sub-networks that specialize in correcting different error types. In this paper, we propose a mixture-of-experts model, MoECE, for grammatical error correction. Our model successfully achieves the performance of T5-XL with three times fewer effective parameters. Additionally, our model produces interpretable corrections by also identifying the error type during inference.

Timeline summarization (TLS) is essential for distilling coherent narratives from a vast collection of texts, tracing the progression of events and topics over time. Prior research typically focuses on either event or topic timeline summarization, neglecting the potential synergy of these two forms. In this study, we bridge this gap by introducing a novel approach that leverages large language models (LLMs) for generating both event and topic timelines. Our approach diverges from conventional TLS by prioritizing event detection, leveraging LLMs as pseudo-oracles for incremental event clustering and the construction of timelines from a text stream. As a result, it produces a more interpretable pipeline. Empirical evaluation across four TLS benchmarks reveals that our approach outperforms the best prior published approaches, highlighting the potential of LLMs in timeline summarization for real-world applications.

In this work, we study multi-source test-time model adaptation from user feedback, where K distinct models are established for adaptation. To allow efficient adaptation, we cast the problem as a stochastic decision-making process, aiming to determine the best adapted model after adaptation. We discuss two frameworks: multi-armed bandit learning and multi-armed dueling bandits. Compared to multi-armed bandit learning, the dueling framework allows pairwise collaboration among K models, which is solved by a novel method named Co-UCB proposed in this work. Experiments on six datasets of extractive question answering (QA) show that the dueling framework using Co-UCB is more effective than other strong baselines for our studied problem.

Reasoning about time is of fundamental importance. Many facts are time-dependent. For example, athletes change teams from time to time, and different government officials are elected periodically. Previous time-dependent question answering (QA) datasets tend to be biased in either their coverage of time spans or question types. In this paper, we introduce a comprehensive probing dataset TempReason to evaluate the temporal reasoning capability of large language models. Our dataset includes questions of three temporal reasoning levels. In addition, we also propose a novel learning framework to improve the temporal reasoning capability of large language models, based on temporal span extraction and time-sensitive reinforcement learning. We conducted experiments in closed book QA, open book QA, and reasoning QA settings and demonstrated the effectiveness of our approach.

We advocate the importance of exposing uncertainty on results of language model prompting which display bias modes resembling cognitive biases, and propose to help users grasp the level of uncertainty via simple quantifying metrics. Cognitive biases in the human decision making process can lead to flawed responses when we are under uncertainty. Not surprisingly, we have seen biases in language models resembling cognitive biases as a result of training on biased textual data, raising dangers in downstream tasks that are centered around people’s lives if users trust their results too much. In this work, we reveal two bias modes leveraging cognitive biases when we prompt BERT, accompanied by two bias metrics. On a drug-drug interaction extraction task, our bias measurements reveal an error pattern similar to the availability bias when the labels for training prompts are imbalanced, and show that a toning-down transformation of the drug-drug description in a prompt can elicit a bias similar to the framing effect, warning users to distrust when prompting language models for answers.

Relation extraction (RE) aims to extract relations from sentences and documents. Existing relation extraction models typically rely on supervised machine learning. However, recent studies showed that many RE datasets are incompletely annotated. This is known as the false negative problem in which valid relations are falsely annotated as ‘no_relation’. Models trained with such data inevitably make similar mistakes during the inference stage. Self-training has been proven effective in alleviating the false negative problem. However, traditional self-training is vulnerable to confirmation bias and exhibits poor performance in minority classes. To overcome this limitation, we proposed a novel class-adaptive re-sampling self-training framework. Specifically, we re-sampled the pseudo-labels for each class by precision and recall scores. Our re-sampling strategy favored the pseudo-labels of classes with high precision and low recall, which improved the overall recall without significantly compromising precision. We conducted experiments on document-level and biomedical relation extraction datasets, and the results showed that our proposed self-training framework consistently outperforms existing competitive methods on the Re-DocRED and ChemDisgene datasets when the training data are incompletely annotated.

State-of-the-art grammatical error correction (GEC) systems rely on parallel training data (ungrammatical sentences and their manually corrected counterparts), which are expensive to construct. In this paper, we employ the Break-It-Fix-It (BIFI) method to build an unsupervised GEC system. The BIFI framework generates parallel data from unlabeled text using a fixer to transform ungrammatical sentences into grammatical ones, and a critic to predict sentence grammaticality. We present an unsupervised approach to build the fixer and the critic, and an algorithm that allows them to iteratively improve each other. We evaluate our unsupervised GEC system on English and Chinese GEC. Empirical results show that our GEC system outperforms previous unsupervised GEC systems, and achieves performance comparable to supervised GEC systems without ensemble. Furthermore, when combined with labeled training data, our system achieves new state-of-the-art results on the CoNLL-2014 and NLPCC-2018 test sets.

Quality estimation models have been developed to assess the corrections made by grammatical error correction (GEC) models when the reference or gold-standard corrections are not available. An ideal quality estimator can be utilized to combine the outputs of multiple GEC systems by choosing the best subset of edits from the union of all edits proposed by the GEC base systems. However, we found that existing GEC quality estimation models are not good enough in differentiating good corrections from bad ones, resulting in a low F0.5 score when used for system combination. In this paper, we propose GRECO, a new state-of-the-art quality estimation model that gives a better estimate of the quality of a corrected sentence, as indicated by having a higher correlation to the F0.5 score of a corrected sentence. It results in a combined GEC system with a higher F0.5 score. We also propose three methods for utilizing GEC quality estimation models for system combination with varying generality: model-agnostic, model-agnostic with voting bias, and model-dependent method. The combined GEC system outperforms the state of the art on the CoNLL-2014 test set and the BEA-2019 test set, achieving the highest F0.5 scores published to date.

The most popular approach in grammatical error correction (GEC) is based on sequence-to-sequence (seq2seq) models. Similar to other autoregressive generation tasks, seq2seq GEC also faces the exposure bias problem, i.e., the context tokens are drawn from different distributions during training and testing, caused by the teacher forcing mechanism. In this paper, we propose a novel data manipulation approach to overcome this problem, which includes a data augmentation method during training to mimic the decoder input at inference time, and a data reweighting method to automatically balance the importance of each kind of augmented samples. Experimental results on benchmark GEC datasets show that our method achieves significant improvements compared to prior approaches.

In this paper, we present ALLECS, a lightweight web application to serve grammatical error correction (GEC) systems so that they can be easily used by the general public. We design ALLECS to be accessible to as many users as possible, including users who have a slow Internet connection and who use mobile phones as their main devices to connect to the Internet. ALLECS provides three state-of-the-art base GEC systems using two approaches (sequence-to-sequence generation and sequence tagging), as well as two state-of-the-art GEC system combination methods using two approaches (edit-based and text-based). ALLECS can be accessed at https://sterling8.d2.comp.nus.edu.sg/gec-demo/

Grammatical Error Correction (GEC) is the task of automatically detecting and correcting errors in text. The task not only includes the correction of grammatical errors, such as missing prepositions and mismatched subject–verb agreement, but also orthographic and semantic errors, such as misspellings and word choice errors, respectively. The field has seen significant progress in the last decade, motivated in part by a series of five shared tasks, which drove the development of rule-based methods, statistical classifiers, statistical machine translation, and finally neural machine translation systems, which represent the current dominant state of the art. In this survey paper, we condense the field into a single article and first outline some of the linguistic challenges of the task, introduce the most popular datasets that are available to researchers (for both English and other languages), and summarize the various methods and techniques that have been developed with a particular focus on artificial error generation. We next describe the many different approaches to evaluation as well as concerns surrounding metric reliability, especially in relation to subjective human judgments, before concluding with an overview of recent progress and suggestions for future work and remaining challenges. We hope that this survey will serve as a comprehensive resource for researchers who are new to the field or who want to be kept apprised of recent developments.

Times flies and it has been close to five and a half years since I became the editor-in-chief of Computational Linguistics on 15 July 2018. In this editorial, I will describe the changes that I have introduced at the journal, and highlight the achievements and challenges of the journal.

In this paper, we formulate system combination for grammatical error correction (GEC) as a simple machine learning task: binary classification. We demonstrate that with the right problem formulation, a simple logistic regression algorithm can be highly effective for combining GEC models. Our method successfully increases the F0.5 score from the highest base GEC system by 4.2 points on the CoNLL-2014 test set and 7.2 points on the BEA-2019 test set. Furthermore, our method outperforms the state of the art by 4.0 points on the BEA-2019 test set, 1.2 points on the CoNLL-2014 test set with original annotation, and 3.4 points on the CoNLL-2014 test set with alternative annotation. We also show that our system combination generates better corrections with higher F0.5 scores than the conventional ensemble.

In conversational question answering (CQA), the task of question rewriting (QR) in context aims to rewrite a context-dependent question into an equivalent self-contained question that gives the same answer. In this paper, we are interested in the robustness of a QR system to questions varying in rewriting hardness or difficulty. Since there is a lack of questions classified based on their rewriting hardness, we first propose a heuristic method to automatically classify questions into subsets of varying hardness, by measuring the discrepancy between a question and its rewrite. To find out what makes questions hard or easy for rewriting, we then conduct a human evaluation to annotate the rewriting hardness of questions. Finally, to enhance the robustness of QR systems to questions of varying hardness, we propose a novel learning framework for QR that first trains a QR model independently on each subset of questions of a certain level of hardness, then combines these QR models as one joint model for inference. Experimental results on two datasets show that our framework improves the overall performance compared to the baselines.

The success of a natural language processing (NLP) system on a task does not amount to fully understanding the complexity of the task, typified by many deep learning models. One such question is: can a black-box model make logically consistent predictions for transitive relations? Recent studies suggest that pre-trained BERT can capture lexico-semantic clues from words in the context. However, to what extent BERT captures the transitive nature of some lexical relations is unclear. From a probing perspective, we examine WordNet word senses and the IS-A relation, which is a transitive relation. That is, for senses A, B, and C, A is-a B and B is-a C entail A is-a C. We aim to quantify how much BERT agrees with the transitive property of IS-A relations, via a minimalist probing setting. Our investigation reveals that BERT’s predictions do not fully obey the transitivity property of the IS-A relation.

The DocRED dataset is one of the most popular and widely used benchmarks for document-level relation extraction (RE). It adopts a recommend-revise annotation scheme so as to have a large-scale annotated dataset. However, we find that the annotation of DocRED is incomplete, i.e., false negative samples are prevalent. We analyze the causes and effects of the overwhelming false negative problem in the DocRED dataset. To address the shortcoming, we re-annotate 4,053 documents in the DocRED dataset by adding the missed relation triples back to the original DocRED. We name our revised DocRED dataset Re-DocRED. We conduct extensive experiments with state-of-the-art neural models on both datasets, and the experimental results show that the models trained and evaluated on our Re-DocRED achieve performance improvements of around 13 F1 points. Moreover, we conduct a comprehensive analysis to identify the potential areas for further improvement.

Document-level Relation Extraction (DocRE) is a more challenging task compared to its sentence-level counterpart. It aims to extract relations from multiple sentences at once. In this paper, we propose a semi-supervised framework for DocRE with three novel components. Firstly, we use an axial attention module for learning the interdependency among entity-pairs, which improves the performance on two-hop relations. Secondly, we propose an adaptive focal loss to tackle the class imbalance problem of DocRE. Lastly, we use knowledge distillation to overcome the differences between human annotated data and distantly supervised data. We conducted experiments on two DocRE datasets. Our model consistently outperforms strong baselines and its performance exceeds the previous SOTA by 1.36 F1 and 1.46 Ign_F1 score on the DocRED leaderboard.

A deployed question answering (QA) model can easily fail when the test data has a distribution shift compared to the training data. Robustness tuning (RT) methods have been widely studied to enhance model robustness against distribution shifts before model deployment. However, can we improve a model after deployment? To answer this question, we evaluate test-time adaptation (TTA) to improve a model after deployment. We first introduce ColdQA, a unified evaluation benchmark for robust QA against text corruption and changes in language and domain. We then evaluate previous TTA methods on ColdQA and compare them to RT methods. We also propose a novel TTA method called online imitation learning (OIL). Through extensive experiments, we find that TTA is comparable to RT methods, and applying TTA after RT can significantly boost the performance on ColdQA. Our proposed OIL improves TTA to be more robust to variation in hyper-parameters and test distributions over time.

There has been much recent progress in natural language processing, and grammatical error correction (GEC) is no exception. We found that state-of-the-art GEC systems (T5 and GECToR) outperform humans by a wide margin on the CoNLL-2014 test set, a benchmark GEC test corpus, as measured by the standard F0.5 evaluation metric. However, a careful examination of their outputs reveals that there are still classes of errors that they fail to correct. This suggests that creating new test data that more accurately measure the true performance of GEC systems constitutes important future work.

While there is much research on cross-domain text classification, most existing approaches focus on one-to-one or many-to-one domain adaptation. In this paper, we tackle the more challenging task of domain generalization, in which domain-invariant representations are learned from multiple source domains, without access to any data from the target domains, and classification decisions are then made on test documents in unseen target domains. We propose a novel framework based on supervised contrastive learning with a memory-saving queue. In this way, we explicitly encourage examples of the same class to be closer and examples of different classes to be further apart in the embedding space. We have conducted extensive experiments on two Amazon review sentiment datasets, and one rumour detection dataset. Experimental results show that our domain generalization method consistently outperforms state-of-the-art domain adaptation methods.

In this paper, we propose a system combination method for grammatical error correction (GEC), based on nonlinear integer programming (IP). Our method optimizes a novel F score objective based on error types, and combines multiple end-to-end GEC systems. The proposed IP approach optimizes the selection of a single best system for each grammatical error type present in the data. Experiments of the IP approach on combining state-of-the-art standalone GEC systems show that the combined system outperforms all standalone systems. It improves F0.5 score by 3.61% when combining the two best participating systems in the BEA 2019 shared task, and achieves F0.5 score of 73.08%. We also perform experiments to compare our IP approach with another state-of-the-art system combination method for GEC, demonstrating IP’s competitive combination capability.

Distantly supervised datasets for relation extraction mostly focus on sentence-level extraction, and they cover very few relations. In this work, we propose cross-document relation extraction, where the two entities of a relation tuple appear in two different documents that are connected via a chain of common entities. Following this idea, we create a dataset for two-hop relation extraction, where each chain contains exactly two documents. Our proposed dataset covers a higher number of relations than the publicly available sentence-level datasets. We also propose a hierarchical entity graph convolutional network (HEGCN) model for this task that improves performance by 1.1% F1 score on our two-hop relation extraction dataset, compared to some strong neural baselines.

Multi-hop question answering (QA) requires a model to retrieve and integrate information from multiple passages to answer a question. Rapid progress has been made on multi-hop QA systems with regard to standard evaluation metrics, including EM and F1. However, by simply evaluating the correctness of the answers, it is unclear to what extent these systems have learned the ability to perform multi-hop reasoning. In this paper, we propose an additional sub-question evaluation for the multi-hop QA dataset HotpotQA, in order to shed some light on explaining the reasoning process of QA systems in answering complex questions. We adopt a neural decomposition model to generate sub-questions for a multi-hop question, followed by extracting the corresponding sub-answers. Contrary to our expectation, multiple state-of-the-art multi-hop QA models fail to answer a large portion of sub-questions, although the corresponding multi-hop questions are correctly answered. Our work takes a step forward towards building a more explainable multi-hop QA system.

Current state-of-the-art supervised word sense disambiguation (WSD) systems (such as GlossBERT and bi-encoder model) yield surprisingly good results by purely leveraging pre-trained language models and short dictionary definitions (or glosses) of the different word senses. While concise and intuitive, the sense gloss is just one of many ways to provide information about word senses. In this paper, we focus on enhancing the sense representations via incorporating synonyms, example phrases or sentences showing usage of word senses, and sense gloss of hypernyms. We show that incorporating such additional information boosts the performance on WSD. With the proposed enhancements, our system achieves an F1 score of 82.0% on the standard benchmark test dataset of the English all-words WSD task, surpassing all previous published scores on this benchmark dataset.

Although grammatical error correction (GEC) has achieved good performance on texts written by learners of English as a second language, performance on low error density domains where texts are written by English speakers of varying levels of proficiency can still be improved. In this paper, we propose a contrastive learning approach to encourage the GEC model to assign a higher probability to a correct sentence while reducing the probability of incorrect sentences that the model tends to generate, so as to improve the accuracy of the model. Experimental results show that our approach significantly improves the performance of GEC models in low error density domains, when evaluated on the benchmark CWEB dataset.

Syntactic dependency parsing is an important task in natural language processing. Unsupervised dependency parsing aims to learn a dependency parser from sentences that have no annotation of their correct parse trees. Despite its difficulty, unsupervised parsing is an interesting research direction because of its capability of utilizing almost unlimited unannotated text data. It also serves as the basis for other research in low-resource parsing. In this paper, we survey existing approaches to unsupervised dependency parsing, identify two major classes of approaches, and discuss recent trends. We hope that our survey can provide insights for researchers and facilitate future research on this topic.

Ensuring smooth communication is essential in a chat-oriented dialogue system, so that a user can obtain meaningful responses through interactions with the system. Most prior work on dialogue research does not focus on preventing dialogue breakdown. One of the major challenges is that a dialogue system may generate an undesired utterance leading to a dialogue breakdown, which degrades the overall interaction quality. Hence, it is crucial for a machine to detect dialogue breakdowns in an ongoing conversation. In this paper, we propose a novel dialogue breakdown detection model that jointly incorporates a pretrained cross-lingual language model and a co-attention network. Our proposed model leverages effective word embeddings trained on one hundred different languages to generate contextualized representations. Co-attention aims to capture the interaction between the latest utterance and the conversation history, and thereby determines whether the latest utterance causes a dialogue breakdown. Experimental results show that our proposed model outperforms all previous approaches on all evaluation metrics in both the Japanese and English tracks in Dialogue Breakdown Detection Challenge 4 (DBDC4 at IWSDS2019).

Despite recent progress in conversational question answering, most prior work does not focus on follow-up questions. Practical conversational question answering systems often receive follow-up questions in an ongoing conversation, and it is crucial for a system to be able to determine whether a question is a follow-up question of the current conversation, for more effective answer finding subsequently. In this paper, we introduce a new follow-up question identification task. We propose a three-way attentive pooling network that determines the suitability of a follow-up question by capturing pair-wise interactions between the associated passage, the conversation history, and a candidate follow-up question. It enables the model to capture topic continuity and topic shift while scoring a particular candidate follow-up question. Experiments show that our proposed three-way attentive pooling network outperforms all baseline systems by significant margins.

Adapting pre-trained language models (PrLMs) (e.g., BERT) to new domains has gained much attention recently. Instead of fine-tuning PrLMs as done in most previous work, we investigate how to adapt the features of PrLMs to new domains without fine-tuning. We explore unsupervised domain adaptation (UDA) in this paper. With the features from PrLMs, we adapt the models trained with labeled data from the source domain to the unlabeled target domain. Self-training is widely used for UDA, and it predicts pseudo labels on the target domain data for training. However, the predicted pseudo labels inevitably include noise, which will negatively affect training a robust model. To improve the robustness of self-training, in this paper we present class-aware feature self-distillation (CFd) to learn discriminative features from PrLMs, in which PrLM features are self-distilled into a feature adaptation module and the features from the same class are more tightly clustered. We further extend CFd to a cross-language setting, in which language discrepancy is studied. Experiments on two monolingual and multilingual Amazon review datasets show that CFd can consistently improve the performance of self-training in cross-domain and cross-language settings.

Automatic grammatical error correction (GEC) research has made remarkable progress in the past decade. However, all existing approaches to GEC correct errors by considering a single sentence alone and ignoring crucial cross-sentence context. Some errors can only be corrected reliably using cross-sentence context and models can also benefit from the additional contextual information in correcting other errors. In this paper, we address this serious limitation of existing approaches and improve strong neural encoder-decoder models by appropriately modeling wider contexts. We employ an auxiliary encoder that encodes previous sentences and incorporate the encoding in the decoder via attention and gating mechanisms. Our approach results in statistically significant improvements in overall GEC performance over strong baselines across multiple test sets. Analysis of our cross-sentence GEC model on a synthetic dataset shows high performance in verb tense corrections that require cross-sentence context.

Aspect-based sentiment analysis produces a list of aspect terms and their corresponding sentiments for a natural language sentence. This task is usually done in a pipeline manner, with aspect term extraction performed first, followed by sentiment predictions toward the extracted aspect terms. While easier to develop, such an approach does not fully exploit joint information from the two subtasks and does not use all available sources of training information that might be helpful, such as document-level labeled sentiment corpus. In this paper, we propose an interactive multi-task learning network (IMN) which is able to jointly learn multiple related tasks simultaneously at both the token level as well as the document level. Unlike conventional multi-task learning methods that rely on learning common features for the different tasks, IMN introduces a message passing architecture where information is iteratively passed to different tasks through a shared set of latent variables. Experimental results demonstrate superior performance of the proposed method against multiple baselines on three benchmark datasets.

Despite the advancement of question answering (QA) systems and rapid improvements on held-out test sets, their generalizability is a topic of concern. We explore the robustness of QA models to question paraphrasing by creating two test sets consisting of paraphrased SQuAD questions. Paraphrased questions from the first test set are very similar to the original questions designed to test QA models’ over-sensitivity, while questions from the second test set are paraphrased using context words near an incorrect answer candidate in an attempt to confuse QA models. We show that both paraphrased test sets lead to significant decrease in performance on multiple state-of-the-art QA models. Using a neural paraphrasing model trained to generate multiple paraphrased questions for a given source question and a set of paraphrase suggestions, we propose a data augmentation approach that requires no human intervention to re-train the models for improved robustness to question paraphrasing.

Relation extraction is the task of determining the relation between two entities in a sentence. Distantly-supervised models are popular for this task. However, sentences can be long and two entities can be located far from each other in a sentence. The pieces of evidence supporting the presence of a relation between two entities may not be very direct, since the entities may be connected via some indirect links such as a third entity or via co-reference. Relation extraction in such scenarios becomes more challenging as we need to capture the long-distance interactions among the entities and other words in the sentence. Also, the words in a sentence do not contribute equally in identifying the relation between the two entities. To address this issue, we propose a novel and effective attention model which incorporates syntactic information of the sentence and a multi-factor attention mechanism. Experiments on the New York Times corpus show that our proposed model outperforms prior state-of-the-art models.

Contextualized word representations are able to give different representations for the same word in different contexts, and they have been shown to be effective in downstream natural language processing tasks, such as question answering, named entity recognition, and sentiment analysis. However, evaluation on word sense disambiguation (WSD) in prior work shows that using contextualized word representations does not outperform the state-of-the-art approach that makes use of non-contextualized word embeddings. In this paper, we explore different strategies of integrating pre-trained contextualized word representations and our best strategy achieves accuracies exceeding the best prior published accuracies by significant margins on multiple benchmark WSD datasets.

The objective of this work is to develop an automated diagnosis system that is able to predict the probability of appendicitis given a free-text emergency department (ED) note and additional structured information (e.g., lab test results). Our clinical corpus consists of about 180,000 ED notes based on ten years of patient visits to the Accident and Emergency (A&E) Department of the National University Hospital (NUH), Singapore. We propose a novel neural network approach that learns to diagnose acute appendicitis based on doctors’ free-text ED notes without any feature engineering. On a test set of 2,000 ED notes with equal number of appendicitis (positive) and non-appendicitis (negative) diagnosis and in which all the negative ED notes only consist of abdominal-related diagnosis, our model is able to achieve a promising F_0.5-score of 0.895 while ED doctors achieve F_0.5-score of 0.900. Visualization shows that our model is able to learn important features, signs, and symptoms of patients from unstructured free-text ED notes, which will help doctors to make better diagnosis.

Attention-based long short-term memory (LSTM) networks have proven to be useful in aspect-level sentiment classification. However, due to the difficulties in annotating aspect-level data, existing public datasets for this task are all relatively small, which largely limits the effectiveness of those neural models. In this paper, we explore two approaches that transfer knowledge from document-level data, which is much less expensive to obtain, to improve the performance of aspect-level sentiment classification. We demonstrate the effectiveness of our approaches on 4 public datasets from SemEval 2014, 2015, and 2016, and we show that attention-based LSTM benefits from document-level knowledge in multiple ways.

Aspect-level sentiment classification aims to determine the sentiment polarity of a review sentence towards an opinion target. A sentence could contain multiple sentiment-target pairs; thus the main challenge of this task is to separate different opinion contexts for different targets. To this end, attention mechanism has played an important role in previous state-of-the-art neural models. The mechanism is able to capture the importance of each context word towards a target by modeling their semantic associations. We build upon this line of research and propose two novel approaches for improving the effectiveness of attention. First, we propose a method for target representation that better captures the semantic meaning of the opinion target. Second, we introduce an attention model that incorporates syntactic information into the attention mechanism. We experiment on attention-based LSTM (Long Short-Term Memory) models using the datasets from SemEval 2014, 2015, and 2016. The experimental results show that the conventional attention-based LSTM can be substantially improved by incorporating the two approaches.

Several metrics have been proposed for evaluating grammatical error correction (GEC) systems based on grammaticality, fluency, and adequacy of the output sentences. Previous studies of the correlation of these metrics with human quality judgments were inconclusive, due to the lack of appropriate significance tests, discrepancies in the methods, and choice of datasets used. In this paper, we re-evaluate reference-based GEC metrics by measuring the system-level correlations with humans on a large dataset of human judgments of GEC outputs, and by properly conducting statistical significance tests. Our results show no significant advantage of GLEU over MaxMatch (M2), contradicting previous studies that claim GLEU to be superior. For a finer-grained analysis, we additionally evaluate these metrics for their agreement with human judgments at the sentence level. Our sentence-level analysis indicates that comparing GLEU and M2, one metric may be more useful than the other depending on the scenario. We further qualitatively analyze these metrics and our findings show that apart from being less interpretable and non-deterministic, GLEU also produces counter-intuitive scores in commonly occurring test examples.

Grammatical error correction (GEC) systems deployed in language learning environments are expected to accurately correct errors in learners’ writing. However, in practice, they often produce spurious corrections and fail to correct many errors, thereby misleading learners. This necessitates the estimation of the quality of output sentences produced by GEC systems so that instructors can selectively intervene and re-correct the sentences which are poorly corrected by the system and ensure that learners get accurate feedback. We propose the first neural approach to automatic quality estimation of GEC output sentences that does not employ any hand-crafted features. Our system is trained in a supervised manner on learner sentences and corresponding GEC system outputs with quality score labels computed using human-annotated references. Our neural quality estimation models for GEC show significant improvements over a strong feature-based baseline. We also show that a state-of-the-art GEC system can be improved when quality scores are used as features for re-ranking the N-best candidates.

We consider the cross-domain sentiment classification problem, where a sentiment classifier is to be learned from a source domain and to be generalized to a target domain. Our approach explicitly minimizes the distance between the source and the target instances in an embedded feature space. With the difference between source and target minimized, we then exploit additional information from the target domain by consolidating the idea of semi-supervised learning, for which, we jointly employ two regularizations — entropy minimization and self-ensemble bootstrapping — to incorporate the unlabeled target data for classifier refinement. Our experimental results demonstrate that the proposed approach can better leverage unlabeled data from the target domain and achieve substantial improvements over baseline methods in various experimental settings.

Recently, there has been a surge of interest in reading comprehension-based (RC) question answering (QA). However, current approaches suffer from an impractical assumption that every question has a valid answer in the associated passage. A practical QA system must possess the ability to determine whether a valid answer exists in a given text passage. In this paper, we focus on developing QA systems that can extract an answer for a question if and only if the associated passage contains an answer. If the associated passage does not contain any valid answer, the QA system will correctly return Nil. We propose a novel nil-aware answer span extraction framework that is capable of returning Nil or a text span from the associated passage as an answer in a single step. We show that our proposed framework can be easily integrated with several recently proposed QA models developed for reading comprehension and can be trained in an end-to-end fashion. Our proposed nil-aware answer extraction neural network decomposes pieces of evidence into relevant and irrelevant parts and then combines them to infer the existence of any answer. Experiments on the NewsQA dataset show that the integration of our proposed framework significantly outperforms several strong baseline systems that use pipeline or threshold-based approaches.

We build a grammatical error correction (GEC) system primarily based on the state-of-the-art statistical machine translation (SMT) approach, using task-specific features and tuning, and further enhance it with the modeling power of neural network joint models. The SMT-based system is weak in generalizing beyond patterns seen during training and lacks granularity below the word level. To address this issue, we incorporate a character-level SMT component targeting the misspelled words that the original SMT-based system fails to correct. Our final system achieves 53.14% F 0.5 score on the benchmark CoNLL-2014 test set, an improvement of 3.62% F 0.5 over the best previous published score.

Aspect extraction is an important and challenging task in aspect-based sentiment analysis. Existing works tend to apply variants of topic models on this task. While fairly successful, these methods usually do not produce highly coherent aspects. In this paper, we present a novel neural approach with the aim of discovering coherent aspects. The model improves coherence by exploiting the distribution of word co-occurrences through the use of neural word embeddings. Unlike topic models which typically assume independently generated words, word embedding models encourage words that appear in similar contexts to be located close to each other in the embedding space. In addition, we use an attention mechanism to de-emphasize irrelevant words during training, further improving the coherence of aspects. Experimental results on real-life datasets demonstrate that our approach discovers more meaningful and coherent aspects, and substantially outperforms baseline methods on several evaluation tasks.

The issue of privacy has always been a concern when clinical texts are used for research purposes. Personal health information (PHI) (such as name and identification number) needs to be removed so that patients cannot be identified. Manual anonymization is not feasible due to the large number of clinical texts to be anonymized. In this paper, we tackle the task of anonymizing clinical texts written in sentence fragments and which frequently contain symbols, abbreviations, and misspelled words. Our clinical texts therefore differ from those in the i2b2 shared tasks which are in prose form with complete sentences. Our clinical texts are also part of a structured database which contains patient name and identification number in structured fields. As such, we formulate our anonymization task as spelling variant detection, exploiting patients’ personal information in the structured fields to detect their spelling variants in clinical texts. We successfully anonymized clinical texts consisting of more than 200 million words, using minimum edit distance and regular expression patterns.

We describe the system developed by the team of the National University of Singapore for the Chinese-English BTEC task of the IWSLT 2009 evaluation campaign. We adopted a state-of-the-art phrase-based statistical machine translation approach and focused on experiments with different Chinese word segmentation standards. In our official submission, we trained a separate system for each segmenter and we combined the outputs in a subsequent re-ranking step. Given the small size of the training data, we further re-trained the system on the development data after tuning. The evaluation results show that both strategies yield sizeable and consistent improvements in translation quality.