COLING 2020 论文列表

Classifiers trained on auxiliary probing tasks are a popular tool to analyze the representations learned by neural sentence encoders such as BERT and ELMo. While many authors are aware of the difficulty to distinguish between “extracting the linguistic structure encoded in the representations” and “learning the probing task,” the validity of probing methods calls for further research. Using a neighboring word identity prediction task, we show that the token embeddings learned by neural sentence encoders contain a significant amount of information about the exact linear context of the token, and hypothesize that, with such information, learning standard probing tasks may be feasible even without additional linguistic structure. We develop this hypothesis into a framework in which analysis efforts can be scrutinized and argue that, with current models and baselines, conclusions that representations contain linguistic structure are not well-founded. Current probing methodology, such as restricting the classifier’s expressiveness or using strong baselines, can help to better estimate the complexity of learning, but not build a foundation for speculations about the nature of the linguistic structure encoded in the learned representations.

Discourse structure tree construction is the fundamental task of discourse parsing and most previous work focused on English. Due to the cultural and linguistic differences, existing successful methods on English discourse parsing cannot be transformed into Chinese directly, especially in paragraph level suffering from longer discourse units and fewer explicit connectives. To alleviate the above issues, we propose two reading modes, i.e., the global backward reading and the local reverse reading, to construct Chinese paragraph level discourse trees. The former processes discourse units from the end to the beginning in a document to utilize the left-branching bias of discourse structure in Chinese, while the latter reverses the position of paragraphs in a discourse unit to enhance the differentiation of coherence between adjacent discourse units. The experimental results on Chinese MCDTB demonstrate that our model outperforms all strong baselines.

Due to the compelling improvements brought by BERT, many recent representation models adopted the Transformer architecture as their main building block, consequently inheriting the wordpiece tokenization system despite it not being intrinsically linked to the notion of Transformers. While this system is thought to achieve a good balance between the flexibility of characters and the efficiency of full words, using predefined wordpiece vocabularies from the general domain is not always suitable, especially when building models for specialized domains (e.g., the medical domain). Moreover, adopting a wordpiece tokenization shifts the focus from the word level to the subword level, making the models conceptually more complex and arguably less convenient in practice. For these reasons, we propose CharacterBERT, a new variant of BERT that drops the wordpiece system altogether and uses a Character-CNN module instead to represent entire words by consulting their characters. We show that this new model improves the performance of BERT on a variety of medical domain tasks while at the same time producing robust, word-level, and open-vocabulary representations.

Most pre-trained language models (PLMs) construct word representations at subword level with Byte-Pair Encoding (BPE) or its variations, by which OOV (out-of-vocab) words are almost avoidable. However, those methods split a word into subword units and make the representation incomplete and fragile. In this paper, we propose a character-aware pre-trained language model named CharBERT improving on the previous methods (such as BERT, RoBERTa) to tackle these problems. We first construct the contextual word embedding for each token from the sequential character representations, then fuse the representations of characters and the subword representations by a novel heterogeneous interaction module. We also propose a new pre-training task named NLM (Noisy LM) for unsupervised character representation learning. We evaluate our method on question answering, sequence labeling, and text classification tasks, both on the original datasets and adversarial misspelling test sets. The experimental results show that our method can significantly improve the performance and robustness of PLMs simultaneously.

To advance understanding on how to engage readers, we advocate the novel task of automatic pull quote selection. Pull quotes are a component of articles specifically designed to catch the attention of readers with spans of text selected from the article and given more salient presentation. This task differs from related tasks such as summarization and clickbait identification by several aspects. We establish a spectrum of baseline approaches to the task, ranging from handcrafted features to a neural mixture-of-experts to cross-task models. By examining the contributions of individual features and embedding dimensions from these models, we uncover unexpected properties of pull quotes to help answer the important question of what engages readers. Human evaluation also supports the uniqueness of this task and the suitability of our selection models. The benefits of exploring this problem further are clear: pull quotes increase enjoyment and readability, shape reader perceptions, and facilitate learning. Code to reproduce this work is available at https://github.com/tannerbohn/AutomaticPullQuoteSelection.

The advent of natural language understanding (NLU) benchmarks for English, such as GLUE and SuperGLUE allows new NLU models to be evaluated across a diverse set of tasks. These comprehensive benchmarks have facilitated a broad range of research and applications in natural language processing (NLP). The problem, however, is that most such benchmarks are limited to English, which has made it difficult to replicate many of the successes in English NLU for other languages. To help remedy this issue, we introduce the first large-scale Chinese Language Understanding Evaluation (CLUE) benchmark. CLUE is an open-ended, community-driven project that brings together 9 tasks spanning several well-established single-sentence/sentence-pair classification tasks, as well as machine reading comprehension, all on original Chinese text. To establish results on these tasks, we report scores using an exhaustive set of current state-of-the-art pre-trained Chinese models (9 in total). We also introduce a number of supplementary datasets and additional tools to help facilitate further progress on Chinese NLU. Our benchmark is released at https://www.cluebenchmarks.com

We introduce CHIME, a cross-passage hierarchical memory network for question answering (QA) via text generation. It extends XLNet introducing an auxiliary memory module consisting of two components: the context memory collecting cross-passage evidences, and the answer memory working as a buffer continually refining the generated answers. Empirically, we show the efficacy of the proposed architecture in the multi-passage generative QA, outperforming the state-of-the-art baselines with better syntactically well-formed answers and increased precision in addressing the questions of the AmazonQA review dataset. An additional qualitative analysis revealed the interpretability introduced by the memory module.

We present the first large scale corpus for entity resolution in email conversations (CEREC). The corpus consists of 6001 email threads from the Enron Email Corpus containing 36,448 email messages and 38,996 entity coreference chains. The annotation is carried out as a two-step process with minimal manual effort. Experiments are carried out for evaluating different features and performance of four baselines on the created corpus. For the task of mention identification and coreference resolution, a best performance of 54.1 F1 is reported, highlighting the room for improvement. An in-depth qualitative and quantitative error analysis is presented to understand the limitations of the baselines considered.

Social media has become the new playground for bullies. Young people are now regularly exposed to a wide range of abuse online. In response to the increasing prevalence of cyberbullying, online social networks have increased efforts to clamp down on online abuse but unfortunately, the nature, complexity and sheer volume of cyberbullying means that many cyberbullying incidents go undetected. BullStop is a mobile app for detecting and preventing cyberbullying and online abuse on social media platforms. It uses deep learning models to identify instances of cyberbullying and can automatically initiate actions such as deleting offensive messages and blocking bullies on behalf of the user. Our system not only achieves impressive prediction results but also demonstrates excellent potential for use in real-world scenarios and is freely available on the Google Play Store.

This paper introduces the first corpus for Automatic Post-Editing of English and a low-resource language, Brazilian Portuguese. The source English texts were extracted from the WebNLG corpus and automatically translated into Portuguese using a state-of-the-art industrial neural machine translator. Post-edits were then obtained in an experiment with native speakers of Brazilian Portuguese. To assess the quality of the corpus, we performed error analysis and computed complexity indicators measuring how difficult the APE task would be. We report preliminary results of Phrase-Based and Neural Machine Translation Models on this new corpus. Data and code publicly available in our repository.

We release large-scale datasets of users’ comments in two languages, English and Korean, for aspect-level sentiment analysis in automotive domain. The datasets consist of 58,000+ commentaspect pairs, which are the largest compared to existing datasets. In addition, this work covers new language (i.e., Korean) along with English for aspect-level sentiment analysis. We build the datasets from automotive domain to enable users (e.g., marketers in automotive companies) to analyze the voice of customers on automobiles. We also provide baseline performances for future work by evaluating recent models on the released datasets.

Named entities pose a unique challenge to traditional methods of language modeling. While several domains are characterised with a high proportion of named entities, the occurrence of specific entities varies widely. Cooking recipes, for example, contain a lot of named entities — viz. ingredients, cooking techniques (also called processes), and utensils. However, some ingredients occur frequently within the instructions while most occur rarely. In this paper, we build upon the previous work done on language models developed for text with named entities by introducing a Hierarchically Disentangled Model. Training is divided into multiple branches with each branch producing a model with overlapping subsets of vocabulary. We found the existing datasets insufficient to accurately judge the performance of the model. Hence, we have curated 158,473 cooking recipes from several publicly available online sources. To reliably derive the entities within this corpus, we employ a combination of Named Entity Recognition (NER) as well as an unsupervised method of interpretation using dependency parsing and POS tagging, followed by a further cleaning of the dataset. This unsupervised interpretation models instructions as action graphs and is specific to the corpus of cooking recipes, unlike NER which is a general method applicable to all corpora. To delve into the utility of our language model, we apply it to tasks such as graph-to-text generation and ingredients-to-recipe generation, comparing it to previous state-of-the-art baselines. We make our dataset (including annotations and processed action graphs) available for use, considering their potential use cases for language modeling and text generation research.

Recent research indicates that taking advantage of complex syntactic features leads to favorable results in Semantic Role Labeling. Nonetheless, an analysis of the latest state-of-the-art multilingual systems reveals the difficulty of bridging the wide gap in performance between high-resource (e.g., English) and low-resource (e.g., German) settings. To overcome this issue, we propose a fully language-agnostic model that does away with morphological and syntactic features to achieve robustness across languages. Our approach outperforms the state of the art in all the languages of the CoNLL-2009 benchmark dataset, especially whenever a scarce amount of training data is available. Our objective is not to reject approaches that rely on syntax, rather to set a strong and consistent language-independent baseline for future innovations in Semantic Role Labeling. We release our model code and checkpoints at https://github.com/SapienzaNLP/multi-srl.

Current supervised relational triple extraction approaches require huge amounts of labeled data and thus suffer from poor performance in few-shot settings. However, people can grasp new knowledge by learning a few instances. To this end, we take the first step to study the few-shot relational triple extraction, which has not been well understood. Unlike previous single-task few-shot problems, relational triple extraction is more challenging as the entities and relations have implicit correlations. In this paper, We propose a novel multi-prototype embedding network model to jointly extract the composition of relational triples, namely, entity pairs and corresponding relations. To be specific, we design a hybrid prototypical learning mechanism that bridges text and knowledge concerning both entities and relations. Thus, implicit correlations between entities and relations are injected. Additionally, we propose a prototype-aware regularization to learn more representative prototypes. Experimental results demonstrate that the proposed method can improve the performance of the few-shot triple extraction.

Bridging reference resolution is an anaphora resolution task that is arguably more challenging and less studied than entity coreference resolution. Given that significant progress has been made on coreference resolution in recent years, we believe that bridging resolution will receive increasing attention in the NLP community. Nevertheless, progress on bridging resolution is currently hampered in part by the scarcity of large annotated corpora for model training as well as the lack of standardized evaluation protocols. This paper presents a survey of the current state of research on bridging reference resolution and discusses future research directions.

Recent image captioning models have made much progress for exploring the multi-modal interaction, such as attention mechanisms. Though these mechanisms can boost the interaction, there are still two gaps between the visual and language domains: (1) the gap between the visual features and textual semantics, (2) the gap between the disordering of visual features and the ordering of texts. To bridge the gaps we propose a high-level semantic planning (HSP) mechanism that incorporates both a semantic reconstruction and an explicit order planning. We integrate the planning mechanism to the attention based caption model and propose the High-level Semantic PLanning based Attention Network (HS-PLAN). First, an attention based reconstruction module is designed to reconstruct the visual features with high-level semantic information. Then we apply a pointer network to serialize the features and obtain the explicit order plan to guide the generation. Experiments conducted on MS COCO show that our model outperforms previous methods and achieves the state-of-the-art performance of 133.4% CIDEr-D score.

In automatic speech translation (ST), traditional cascade approaches involving separate transcription and translation steps are giving ground to increasingly competitive and more robust direct solutions. In particular, by translating speech audio data without intermediate transcription, direct ST models are able to leverage and preserve essential information present in the input (e.g.speaker’s vocal characteristics) that is otherwise lost in the cascade framework. Although such ability proved to be useful for gender translation, direct ST is nonetheless affected by gender bias just like its cascade counterpart, as well as machine translation and numerous other natural language processing applications. Moreover, direct ST systems that exclusively rely on vocal biometric features as a gender cue can be unsuitable or even potentially problematic for certain users. Going beyond speech signals, in this paper we compare different approaches to inform direct ST models about the speaker’s gender and test their ability to handle gender translation from English into Italian and French. To this aim, we manually annotated large datasets with speak-ers’ gender information and used them for experiments reflecting different possible real-world scenarios. Our results show that gender-aware direct ST solutions can significantly outperform strong – but gender-unaware – direct ST models. In particular, the translation of gender-marked words can increase up to 30 points in accuracy while preserving overall translation quality.

We present a bracketing-based encoding that can be used to represent any 2-planar dependency tree over a sentence of length n as a sequence of n labels, hence providing almost total coverage of crossing arcs in sequence labeling parsing. First, we show that existing bracketing encodings for parsing as labeling can only handle a very mild extension of projective trees. Second, we overcome this limitation by taking into account the well-known property of 2-planarity, which is present in the vast majority of dependency syntactic structures in treebanks, i.e., the arcs of a dependency tree can be split into two planes such that arcs in a given plane do not cross. We take advantage of this property to design a method that balances the brackets and that encodes the arcs belonging to each of those planes, allowing for almost unrestricted non-projectivity (∼99.9% coverage) in sequence labeling parsing. The experiments show that our linearizations improve over the accuracy of the original bracketing encoding in highly non-projective treebanks (on average by 0.4 LAS), while achieving a similar speed. Also, they are especially suitable when PoS tags are not used as input parameters to the models.

A promising application of AI to healthcare is the retrieval of information from electronic health records (EHRs), e.g. to aid clinicians in finding relevant information for a consultation or to recruit suitable patients for a study. This requires search capabilities far beyond simple string matching, including the retrieval of concepts (diagnoses, symptoms, medications, etc.) related to the one in question. The suitability of AI methods for such applications is tested by predicting the relatedness of concepts with known relatedness scores. However, all existing biomedical concept relatedness datasets are notoriously small and consist of hand-picked concept pairs. We open-source a novel concept relatedness benchmark overcoming these issues: it is six times larger than existing datasets and concept pairs are chosen based on co-occurrence in EHRs, ensuring their relevance for the application of interest. We present an in-depth analysis of our new dataset and compare it to existing ones, highlighting that it is not only larger but also complements existing datasets in terms of the types of concepts included. Initial experiments with state-of-the-art embedding methods show that our dataset is a challenging new benchmark for testing concept relatedness models.