EMNLP 2021 论文列表

Measuring the similarity score between a pair of sentences in different languages is the essential requisite for multilingual sentence embedding methods. Predicting the similarity score consists of two sub-tasks, which are monolingual similarity evaluation and multilingual sentence retrieval. However, conventional methods have mainly tackled only one of the sub-tasks and therefore showed biased performances. In this paper, we suggest a novel and strong method for multilingual sentence embedding, which shows performance improvement on both sub-tasks, consequently resulting in robust predictions of multilingual similarity scores. The suggested method consists of two parts: to learn semantic similarity of sentences in the pivot language and then to extend the learned semantic structure to different languages. To align semantic structures across different languages, we introduce a teacher-student network. The teacher network distills the knowledge of the pivot language to different languages of the student network. During the distillation, the parameters of the teacher network are updated with the slow-moving average. Together with the distillation and the parameter updating, the semantic structure of the student network can be directly aligned across different languages while preserving the ability to measure the semantic similarity. Thus, the multilingual training method drives performance improvement on multilingual similarity evaluation. The suggested model achieves the state-of-the-art performance on extended STS 2017 multilingual similarity evaluation as well as two sub-tasks, which are extended STS 2017 monolingual similarity evaluation and Tatoeba multilingual retrieval in 14 languages.

While pre-trained language models have obtained state-of-the-art performance for several natural language understanding tasks, they are quite opaque in terms of their decision-making process. While some recent works focus on rationalizing neural predictions by highlighting salient concepts in the text as justifications or rationales, they rely on thousands of labeled training examples for both task labels as well as annotated rationales for every instance. Such extensive large-scale annotations are infeasible to obtain for many tasks. To this end, we develop a multi-task teacher-student framework based on self-training pre-trained language models with limited task-specific labels and rationales and judicious sample selection to learn from informative pseudo-labeled examples. We study several characteristics of what constitutes a good rationale and demonstrate that the neural model performance can be significantly improved by making it aware of its rationalized predictions, particularly in low-resource settings. Extensive experiments in several benchmark datasets demonstrate the effectiveness of our approach.

As the labeling cost for different modules in task-oriented dialog (ToD) systems is expensive, a major challenge is to train different modules with the least amount of labeled data. Recently, large-scale pre-trained language models, have shown promising results for few-shot learning in ToD. In this paper, we devise a self-training approach to utilize the abundant unlabeled dialog data to further improve state-of-the-art pre-trained models in few-shot learning scenarios for ToD systems. Specifically, we propose a self-training approach that iteratively labels the most confident unlabeled data to train a stronger Student model. Moreover, a new text augmentation technique (GradAug) is proposed to better train the Student by replacing non-crucial tokens using a masked language model. We conduct extensive experiments and present analyses on four downstream tasks in ToD, including intent classification, dialog state tracking, dialog act prediction, and response selection. Empirical results demonstrate that the proposed self-training approach consistently improves state-of-the-art pre-trained models (BERT, ToD-BERT) when only a small number of labeled data are available.

Spoken question answering (SQA) requires fine-grained understanding of both spoken documents and questions for the optimal answer prediction. In this paper, we propose novel training schemes for spoken question answering with a self-supervised training stage and a contrastive representation learning stage. In the self-supervised stage, we propose three auxiliary self-supervised tasks, including utterance restoration, utterance insertion, and question discrimination, and jointly train the model to capture consistency and coherence among speech documents without any additional data or annotations. We then propose to learn noise-invariant utterance representations in a contrastive objective by adopting multiple augmentation strategies, including span deletion and span substitution. Besides, we design a Temporal-Alignment attention to semantically align the speech-text clues in the learned common space and benefit the SQA tasks. By this means, the training schemes can more effectively guide the generation model to predict more proper answers. Experimental results show that our model achieves state-of-the-art results on three SQA benchmarks. Our code will be publicly available after publication.

We tackle the problem of self-training networks for NLU in low-resource environment—few labeled data and lots of unlabeled data. The effectiveness of self-training is a result of increasing the amount of training data while training. Yet it becomes less effective in low-resource settings due to unreliable labels predicted by the teacher model on unlabeled data. Rules of grammar, which describe the grammatical structure of data, have been used in NLU for better explainability. We propose to use rules of grammar in self-training as a more reliable pseudo-labeling mechanism, especially when there are few labeled data. We design an effective algorithm that constructs and expands rules of grammar without human involvement. Then we integrate the constructed rules as a pseudo-labeling mechanism into self-training. There are two possible scenarios regarding data distribution: it is unknown or known in prior to training. We empirically demonstrate that our approach substantially outperforms the state-of-the-art methods in three benchmark datasets for both scenarios.

Despite considerable progress, most machine reading comprehension (MRC) tasks still lack sufficient training data to fully exploit powerful deep neural network models with millions of parameters, and it is laborious, expensive, and time-consuming to create large-scale, high-quality MRC data through crowdsourcing. This paper focuses on generating more training data for MRC tasks by leveraging existing question-answering (QA) data. We first collect a large-scale multi-subject multiple-choice QA dataset for Chinese, ExamQA. We next use incomplete, yet relevant snippets returned by a web search engine as the context for each QA instance to convert it into a weakly-labeled MRC instance. To better use the weakly-labeled data to improve a target MRC task, we evaluate and compare several methods and further propose a self-teaching paradigm. Experimental results show that, upon state-of-the-art MRC baselines, we can obtain +5.1% in accuracy on a multiple-choice Chinese MRC dataset, Cˆ3, and +3.8% in exact match on an extractive Chinese MRC dataset, CMRC 2018, demonstrating the usefulness of the generated QA-based weakly-labeled data for different types of MRC tasks as well as the effectiveness of self-teaching. ExamQA will be available at https://dataset.org/examqa/.

Quality estimation (QE) of machine translation (MT) aims to evaluate the quality of machine-translated sentences without references and is important in practical applications of MT. Training QE models require massive parallel data with hand-crafted quality annotations, which are time-consuming and labor-intensive to obtain. To address the issue of the absence of annotated training data, previous studies attempt to develop unsupervised QE methods. However, very few of them can be applied to both sentence- and word-level QE tasks, and they may suffer from noises in the synthetic data. To reduce the negative impact of noises, we propose a self-supervised method for both sentence- and word-level QE, which performs quality estimation by recovering the masked target words. Experimental results show that our method outperforms previous unsupervised methods on several QE tasks in different language pairs and domains.

Topic models are useful tools for analyzing and interpreting the main underlying themes of large corpora of text. Most topic models rely on word co-occurrence for computing a topic, i.e., a weighted set of words that together represent a high-level semantic concept. In this paper, we propose a new light-weight Self-Supervised Neural Topic Model (SNTM) that learns a rich context by learning a topic representation jointly from three co-occurring words and a document that the triple originates from. Our experimental results indicate that our proposed neural topic model, SNTM, outperforms previously existing topic models in coherence metrics as well as document clustering accuracy. Moreover, apart from the topic coherence and clustering performance, the proposed neural topic model has a number of advantages, namely, being computationally efficient and easy to train.

Contextualised word embeddings is a powerful tool to detect contextual synonyms. However, most of the current state-of-the-art (SOTA) deep learning concept extraction methods remain supervised and underexploit the potential of the context. In this paper, we propose a self-supervised pre-training approach which is able to detect contextual synonyms of concepts being training on the data created by shallow matching. We apply our methodology in the sparse multi-class setting (over 15,000 concepts) to extract phenotype information from electronic health records. We further investigate data augmentation techniques to address the problem of the class sparsity. Our approach achieves a new SOTA for the unsupervised phenotype concept annotation on clinical text on F1 and Recall outperforming the previous SOTA with a gain of up to 4.5 and 4.0 absolute points, respectively. After fine-tuning with as little as 20% of the labelled data, we also outperform BioBERT and ClinicalBERT. The extrinsic evaluation on three ICU benchmarks also shows the benefit of using the phenotypes annotated by our model as features.

Spelling Error Correction (SEC) that requires high-level language understanding is a challenging but useful task. Current SEC approaches normally leverage a pre-training then fine-tuning procedure that treats data equally. By contrast, Curriculum Learning (CL) utilizes training data differently during training and has shown its effectiveness in improving both performance and training efficiency in many other NLP tasks. In NMT, a model’s performance has been shown sensitive to the difficulty of training examples, and CL has been shown effective to address this. In SEC, the data from different language learners are naturally distributed at different difficulty levels (some errors made by beginners are obvious to correct while some made by fluent speakers are hard), and we expect that designing a curriculum correspondingly for model learning may also help its training and bring about better performance. In this paper, we study how to further improve the performance of the state-of-the-art SEC method with CL, and propose a Self-Supervised Curriculum Learning (SSCL) approach. Specifically, we directly use the cross-entropy loss as criteria for: 1) scoring the difficulty of training data, and 2) evaluating the competence of the model. In our approach, CL improves the model training, which in return improves the CL measurement. In our experiments on the SIGHAN 2015 Chinese spelling check task, we show that SSCL is superior to previous norm-based and uncertainty-aware approaches, and establish a new state of the art (74.38% F1).

Event detection (ED) task aims to classify events by identifying key event trigger words embedded in a piece of text. Previous research have proved the validity of fusing syntactic dependency relations into Graph Convolutional Networks(GCN). While existing GCN-based methods explore latent node-to-node dependency relations according to a stationary adjacency tensor, an attention-based dynamic tensor, which can pay much attention to the key node like event trigger or its neighboring nodes, has not been developed. Simultaneously, suffering from the phenomenon of graph information vanishing caused by the symmetric adjacency tensor, existing GCN models can not achieve higher overall performance. In this paper, we propose a novel model Self-Attention Graph Residual Convolution Networks (SA-GRCN) to mine node-to-node latent dependency relations via self-attention mechanism and introduce Graph Residual Network (GResNet) to solve graph information vanishing problem. Specifically, a self-attention module is constructed to generate an attention tensor, representing the dependency attention scores of all words in the sentence. Furthermore, a graph residual term is added to the baseline SA-GCN to construct a GResNet. Considering the syntactically connection of the network input, we initialize the raw adjacency tensor without processed by the self-attention module as the residual term. We conduct experiments on the ACE2005 dataset and the results show significant improvement over competitive baseline methods.

Multi-party dialogue machine reading comprehension (MRC) brings tremendous challenge since it involves multiple speakers at one dialogue, resulting in intricate speaker information flows and noisy dialogue contexts. To alleviate such difficulties, previous models focus on how to incorporate these information using complex graph-based modules and additional manually labeled data, which is usually rare in real scenarios. In this paper, we design two labour-free self- and pseudo-self-supervised prediction tasks on speaker and key-utterance to implicitly model the speaker information flows, and capture salient clues in a long dialogue. Experimental results on two benchmark datasets have justified the effectiveness of our method over competitive baselines and current state-of-the-art models.

The pivot for the unified Aspect-based Sentiment Analysis (ABSA) is to couple aspect terms with their corresponding opinion terms, which might further derive easier sentiment predictions. In this paper, we investigate the unified ABSA task from the perspective of Machine Reading Comprehension (MRC) by observing that the aspect and the opinion terms can serve as the query and answer in MRC interchangeably. We propose a new paradigm named Role Flipped Machine Reading Comprehension (RF-MRC) to resolve. At its heart, the predicted results of either the Aspect Term Extraction (ATE) or the Opinion Terms Extraction (OTE) are regarded as the queries, respectively, and the matched opinion or aspect terms are considered as answers. The queries and answers can be flipped for multi-hop detection. Finally, every matched aspect-opinion pair is predicted by the sentiment classifier. RF-MRC can solve the ABSA task without any additional data annotation or transformation. Experiments on three widely used benchmarks and a challenging dataset demonstrate the superiority of the proposed framework.

The span-based model enjoys great popularity in recent works of sequence segmentation. However, each of these methods suffers from its own defects, such as invalid predictions. In this work, we introduce a unified span-based model, lexical unit analysis (LUA), that addresses all these matters. Segmenting a lexical unit sequence involves two steps. Firstly, we embed every span by using the representations from a pretraining language model. Secondly, we define a score for every segmentation candidate and apply dynamic programming (DP) to extract the candidate with the maximum score. We have conducted extensive experiments on 3 tasks, (e.g., syntactic chunking), across 7 datasets. LUA has established new state-of-the-art performances on 6 of them. We have achieved even better results through incorporating label correlations.

Recent state-of-the-art (SOTA) effective neural network methods and fine-tuning methods based on pre-trained models (PTM) have been used in Chinese word segmentation (CWS), and they achieve great results. However, previous works focus on training the models with the fixed corpus at every iteration. The intermediate generated information is also valuable. Besides, the robustness of the previous neural methods is limited by the large-scale annotated data. There are a few noises in the annotated corpus. Limited efforts have been made by previous studies to deal with such problems. In this work, we propose a self-supervised CWS approach with a straightforward and effective architecture. First, we train a word segmentation model and use it to generate the segmentation results. Then, we use a revised masked language model (MLM) to evaluate the quality of the segmentation results based on the predictions of the MLM. Finally, we leverage the evaluations to aid the training of the segmenter by improved minimum risk training. Experimental results show that our approach outperforms previous methods on 9 different CWS datasets with single criterion training and multiple criteria training and achieves better robustness.

Aspect-based sentiment analysis (ABSA) task consists of three typical subtasks: aspect term extraction, opinion term extraction, and sentiment polarity classification. These three subtasks are usually performed jointly to save resources and reduce the error propagation in the pipeline. However, most of the existing joint models only focus on the benefits of encoder sharing between subtasks but ignore the difference. Therefore, we propose a joint ABSA model, which not only enjoys the benefits of encoder sharing but also focuses on the difference to improve the effectiveness of the model. In detail, we introduce a dual-encoder design, in which a pair encoder especially focuses on candidate aspect-opinion pair classification, and the original encoder keeps attention on sequence labeling. Empirical results show that our proposed model shows robustness and significantly outperforms the previous state-of-the-art on four benchmark datasets.

This paper presents Self-correcting Encoding (Secoco), a framework that effectively deals with noisy input for robust neural machine translation by introducing self-correcting predictors. Different from previous robust approaches, Secoco enables NMT to explicitly correct noisy inputs and delete specific errors simultaneously with the translation decoding process. Secoco is able to achieve significant improvements over strong baselines on two real-world test sets and a benchmark WMT dataset with good interpretability. We will make our code and dataset publicly available soon.

Recent studies have shown that deep neural network-based models are vulnerable to intentionally crafted adversarial examples, and various methods have been proposed to defend against adversarial word-substitution attacks for neural NLP models. However, there is a lack of systematic study on comparing different defense approaches under the same attacking setting. In this paper, we seek to fill the gap of systematic studies through comprehensive researches on understanding the behavior of neural text classifiers trained by various defense methods under representative adversarial attacks. In addition, we propose an effective method to further improve the robustness of neural text classifiers against such attacks, and achieved the highest accuracy on both clean and adversarial examples on AGNEWS and IMDB datasets by a significant margin. We hope this study could provide useful clues for future research on text adversarial defense. Codes are available at https://github.com/RockyLzy/TextDefender.

Computational models of human language often involve combinatorial problems. For instance, a probabilistic parser may marginalize over exponentially many trees to make predictions. Algorithms for such problems often employ dynamic programming and are not always unique. Finding one with optimal asymptotic runtime can be unintuitive, time-consuming, and error-prone. Our work aims to automate this laborious process. Given an initial correct declarative program, we search for a sequence of semantics-preserving transformations to improve its running time as much as possible. To this end, we describe a set of program transformations, a simple metric for assessing the efficiency of a transformed program, and a heuristic search procedure to improve this metric. We show that in practice, automated search—like the mental search performed by human programmers—can find substantial improvements to the initial program. Empirically, we show that many speed-ups described in the NLP literature could have been discovered automatically by our system.