COLING 2020 论文列表

This paper proposed a new subword segmentation method for neural machine translation, “Bilingual Subword Segmentation,” which tokenizes sentences to minimize the difference between the number of subword units in a sentence and that of its translation. While existing subword segmentation methods tokenize a sentence without considering its translation, the proposed method tokenizes a sentence by using subword units induced from bilingual sentences; this method could be more favorable to machine translation. Evaluations on WAT Asian Scientific Paper Excerpt Corpus (ASPEC) English-to-Japanese and Japanese-to-English translation tasks and WMT14 English-to-German and German-to-English translation tasks show that our bilingual subword segmentation improves the performance of Transformer neural machine translation (up to +0.81 BLEU).

We introduce Biased TextRank, a graph-based content extraction method inspired by the popular TextRank algorithm that ranks text spans according to their importance for language processing tasks and according to their relevance to an input “focus.” Biased TextRank enables focused content extraction for text by modifying the random restarts in the execution of TextRank. The random restart probabilities are assigned based on the relevance of the graph nodes to the focus of the task. We present two applications of Biased TextRank: focused summarization and explanation extraction, and show that our algorithm leads to improved performance on two different datasets by significant ROUGE-N score margins. Much like its predecessor, Biased TextRank is unsupervised, easy to implement and orders of magnitude faster and lighter than current state-of-the-art Natural Language Processing methods for similar tasks.

We propose a novel Bi-directional Cognitive Knowledge Framework (BCKF) for reading comprehension from the perspective of complementary learning systems theory. It aims to simulate two ways of thinking in the brain to answer questions, including reverse thinking and inertial thinking. To validate the effectiveness of our framework, we design a corresponding Bi-directional Cognitive Thinking Network (BCTN) to encode the passage and generate a question (answer) given an answer (question) and decouple the bi-directional knowledge. The model has the ability to reverse reasoning questions which can assist inertial thinking to generate more accurate answers. Competitive improvement is observed in DuReader dataset, confirming our hypothesis that bi-directional knowledge helps the QA task. The novel framework shows an interesting perspective on machine reading comprehension and cognitive science.

Sign Language Translation (SLT) first uses a Sign Language Recognition (SLR) system to extract sign language glosses from videos. Then, a translation system generates spoken language translations from the sign language glosses. This paper focuses on the translation system and introduces the STMC-Transformer which improves on the current state-of-the-art by over 5 and 7 BLEU respectively on gloss-to-text and video-to-text translation of the PHOENIX-Weather 2014T dataset. On the ASLG-PC12 corpus, we report an increase of over 16 BLEU. We also demonstrate the problem in current methods that rely on gloss supervision. The video-to-text translation of our STMC-Transformer outperforms translation of GT glosses. This contradicts previous claims that GT gloss translation acts as an upper bound for SLT performance and reveals that glosses are an inefficient representation of sign language. For future SLT research, we therefore suggest an end-to-end training of the recognition and translation models, or using a different sign language annotation scheme.

Natural language generation (NLG) is a critical component in conversational systems, owing to its role of formulating a correct and natural text response. Traditionally, NLG components have been deployed using template-based solutions. Although neural network solutions recently developed in the research community have been shown to provide several benefits, deployment of such model-based solutions has been challenging due to high latency, correctness issues, and high data needs. In this paper, we present approaches that have helped us deploy data-efficient neural solutions for NLG in conversational systems to production. We describe a family of sampling and modeling techniques to attain production quality with light-weight neural network models using only a fraction of the data that would be necessary otherwise, and show a thorough comparison between each. Our results show that domain complexity dictates the appropriate approach to achieve high data efficiency. Finally, we distill the lessons from our experimental findings into a list of best practices for production-level NLG model development, and present them in a brief runbook. Importantly, the end products of all of the techniques are small sequence-to-sequence models (~2Mb) that we can reliably deploy in production. These models achieve the same quality as large pretrained models (~1Gb) as judged by human raters.

The classic deep learning paradigm learns a model from the training data of a single task and the learned model is also tested on the same task. This paper studies the problem of learning a sequence of tasks (sentiment classification tasks in our case). After each sentiment classification task is learned, its knowledge is retained to help future task learning. Following this setting, we explore attention neural networks and propose a Bayes-enhanced Lifelong Attention Network (BLAN). The key idea is to exploit the generative parameters of naive Bayes to learn attention knowledge. The learned knowledge from each task is stored in a knowledge base and later used to build lifelong attentions. The constructed lifelong attentions are then used to enhance the attention of the network to help new task learning. Experimental results on product reviews from Amazon.com show the effectiveness of the proposed model.

Joint intent detection and slot filling has recently achieved tremendous success in advancing the performance of utterance understanding. However, many joint models still suffer from the robustness problem, especially on noisy inputs or rare/unseen events. To address this issue, we propose a Joint Adversarial Training (JAT) model to improve the robustness of joint intent detection and slot filling, which consists of two parts: (1) automatically generating joint adversarial examples to attack the joint model, and (2) training the model to defend against the joint adversarial examples so as to robustify the model on small perturbations. As the generated joint adversarial examples have different impacts on the intent detection and slot filling loss, we further propose a Balanced Joint Adversarial Training (BJAT) model that applies a balance factor as a regularization term to the final loss function, which yields a stable training procedure. Extensive experiments and analyses on the lightweight models show that our proposed methods achieve significantly higher scores and substantially improve the robustness of both intent detection and slot filling. In addition, the combination of our BJAT with BERT-large achieves state-of-the-art results on two datasets.

The meaning of natural language text is supported by cohesion among various kinds of entities, including coreference relations, predicate-argument structures, and bridging anaphora relations. However, predicate-argument structures for nominal predicates and bridging anaphora relations have not been studied well, and their analyses have been still very difficult. Recent advances in neural networks, in particular, self training-based language models including BERT (Devlin et al., 2019), have significantly improved many natural language processing tasks, making it possible to dive into the study on analysis of cohesion in the whole text. In this study, we tackle an integrated analysis of cohesion in Japanese texts. Our results significantly outperformed existing studies in each task, especially about 10 to 20 point improvement both for zero anaphora and coreference resolution. Furthermore, we also showed that coreference resolution is different in nature from the other tasks and should be treated specially.

This paper proposes an iterative inference algorithm for multi-hop explanation regeneration, that retrieves relevant factual evidence in the form of text snippets, given a natural language question and its answer. Combining multiple sources of evidence or facts for multi-hop reasoning becomes increasingly hard when the number of sources needed to make an inference grows. Our algorithm copes with this by decomposing the selection of facts from a corpus autoregressively, conditioning the next iteration on previously selected facts. This allows us to use a pairwise learning-to-rank loss. We validate our method on datasets of the TextGraphs 2019 and 2020 Shared Tasks for explanation regeneration. Existing work on this task either evaluates facts in isolation or artificially limits the possible chains of facts, thus limiting multi-hop inference. We demonstrate that our algorithm, when used with a pre-trained transformer model, outperforms the previous state-of-the-art in terms of precision, training time and inference efficiency.

Human natural language is mentioned at a specific point in time while human emotions change over time. While much work has established a strong link between language use and emotional states, few have attempted to model emotional language in time. Here, we introduce the task of affective language forecasting – predicting future change in language based on past changes of language, a task with real-world applications such as treating mental health or forecasting trends in consumer confidence. We establish some of the fundamental autoregressive characteristics of the task (necessary history size, static versus dynamic length, varying time-step resolutions) and then build on popular sequence models for words to instead model sequences of language-based emotion in time. Over a novel Twitter dataset of 1,900 users and weekly + daily scores for 6 emotions and 2 additional linguistic attributes, we find a novel dual-sequence GRU model with decayed hidden states achieves best results (r = .66) significantly out-predicting, e.g., a moving averaging based on the past time-steps (r = .49). We make our anonymized dataset as well as task setup and evaluation code available for others to build on.

A recent approach for few-shot text classification is to convert textual inputs to cloze questions that contain some form of task description, process them with a pretrained language model and map the predicted words to labels. Manually defining this mapping between words and labels requires both domain expertise and an understanding of the language model’s abilities. To mitigate this issue, we devise an approach that automatically finds such a mapping given small amounts of training data. For a number of tasks, the mapping found by our approach performs almost as well as hand-crafted label-to-word mappings.

Interlinear Glossed Text (IGT) is a widely used format for encoding linguistic information in language documentation projects and scholarly papers. Manual production of IGT takes time and requires linguistic expertise. We attempt to address this issue by creating automatic glossing models, using modern multi-source neural models that additionally leverage easy-to-collect translations. We further explore cross-lingual transfer and a simple output length control mechanism, further refining our models. Evaluated on three challenging low-resource scenarios, our approach significantly outperforms a recent, state-of-the-art baseline, particularly improving on overall accuracy as well as lemma and tag recall.

To assess knowledge proficiency of a learner, multiple choice question is an efficient and widespread form in standard tests. However, the composition of the multiple choice question, especially the construction of distractors is quite challenging. The distractors are required to both incorrect and plausible enough to confuse the learners who did not master the knowledge. Currently, the distractors are generated by domain experts which are both expensive and time-consuming. This urges the emergence of automatic distractor generation, which can benefit various standard tests in a wide range of domains. In this paper, we propose a question and answer guided distractor generation (EDGE) framework to automate distractor generation. EDGE consists of three major modules: (1) the Reforming Question Module and the Reforming Passage Module apply gate layers to guarantee the inherent incorrectness of the generated distractors; (2) the Distractor Generator Module applies attention mechanism to control the level of plausibility. Experimental results on a large-scale public dataset demonstrate that our model significantly outperforms existing models and achieves a new state-of-the-art.

This paper presents AutoGOAL, a system for automatic machine learning (AutoML) that uses heterogeneous techniques. In contrast with existing AutoML approaches, our contribution can automatically build machine learning pipelines that combine techniques and algorithms from different frameworks, including shallow classifiers, natural language processing tools, and neural networks. We define the heterogeneous AutoML optimization problem as the search for the best sequence of algorithms that transforms specific input data into the desired output. This provides a novel theoretical and practical approach to AutoML. Our proposal is experimentally evaluated in diverse machine learning problems and compared with alternative approaches, showing that it is competitive with other AutoML alternatives in standard benchmarks. Furthermore, it can be applied to novel scenarios, such as several NLP tasks, where existing alternatives cannot be directly deployed. The system is freely available and includes in-built compatibility with a large number of popular machine learning frameworks, which makes our approach useful for solving practical problems with relative ease and effort.

Text generative models (TGMs) excel in producing text that matches the style of human language reasonably well. Such TGMs can be misused by adversaries, e.g., by automatically generating fake news and fake product reviews that can look authentic and fool humans. Detectors that can distinguish text generated by TGM from human written text play a vital role in mitigating such misuse of TGMs. Recently, there has been a flurry of works from both natural language processing (NLP) and machine learning (ML) communities to build accurate detectors for English. Despite the importance of this problem, there is currently no work that surveys this fast-growing literature and introduces newcomers to important research challenges. In this work, we fill this void by providing a critical survey and review of this literature to facilitate a comprehensive understanding of this problem. We conduct an in-depth error analysis of the state-of-the-art detector and discuss research directions to guide future work in this exciting area.

Automatic Charge Identification (ACI) is the task of identifying the relevant charges given the facts of a situation and the statutory laws that define these charges, and is a crucial aspect of the judicial process. Existing works focus on learning charge-side representations by modeling relationships between the charges, but not much effort has been made in improving fact-side representations. We observe that only a small fraction of sentences in the facts actually indicates the charges. We show that by using a very small subset (< 3%) of fact descriptions annotated with sentence-level charges, we can achieve an improvement across a range of different ACI models, as compared to modeling just the main document-level task on a much larger dataset. Additionally, we propose a novel model that utilizes sentence-level charge labels as an auxiliary task, coupled with the main task of document-level charge identification in a multi-task learning framework. The proposed model comprehensively outperforms a large number of recent baselines for ACI. The improvement in performance is particularly noticeable for the rare charges which are known to be especially challenging to identify.

This paper describes a writing assistance system that helps students improve their academic writing. Given an input text, the system suggests lexical substitutions that aim to incorporate more academic vocabulary. The substitution candidates are drawn from an academic word list and ranked by a masked language model. Experimental results show that lexical formality analysis can improve the quality of the suggestions, in comparison to a baseline that relies on the masked language model only.

This paper brings together approaches from the fields of NLP and psychometric measurement to address the problem of predicting examinee proficiency from responses to short-answer questions (SAQs). While previous approaches train on manually labeled data to predict the human-ratings assigned to SAQ responses, the approach presented here models examinee proficiency directly and does not require manually labeled data to train on. We use data from a large medical exam where experimental SAQ items are embedded alongside 106 scored multiple-choice questions (MCQs). First, the latent trait of examinee proficiency is measured using the scored MCQs and then a model is trained on the experimental SAQ responses as input, aiming to predict proficiency as its target variable. The predicted value is then used as a “score” for the SAQ response and evaluated in terms of its contribution to the precision of proficiency estimation.

This paper proposes a novel miscellaneous-context-based method to convert a sentence into a knowledge embedding in the form of a directed graph. We adopt the idea of conceptual graphs to frame for the miscellaneous textual information into conceptual compactness. We first empirically observe that this graph representation method can (1) accommodate the slot-filling challenges in typical question answering and (2) access to the sentence-level graph structure in order to explicitly capture the neighbouring connections of reference concept nodes. Secondly, we propose a task-agnostic semantics-measured module, which cooperates with the graph representation method, in order to (3) project an edge of a sentence-level graph to the space of semantic relevance with respect to the corresponding concept nodes. As a result of experiments on the QA-type relation extraction, the combination of the graph representation and the semantics-measured module achieves the high accuracy of answer prediction and offers human-comprehensible graphical interpretation for every well-formed sample. To our knowledge, our approach is the first towards the interpretable process of learning vocabulary representations with the experimental evidence.