COLING 2020 Papers

We probe the layers in multilingual BERT (mBERT) for phylogenetic and geographic language signals across 100 languages and compute language distances based on the mBERT representations. We 1) employ the language distances to infer and evaluate language trees, finding that they are close to the reference family tree in terms of quartet tree distance, 2) perform distance matrix regression analysis, finding that the language distances can be best explained by phylogenetic and worst by structural factors and 3) present a novel measure for measuring diachronic meaning stability (based on cross-lingual representation variability) which correlates significantly with published ranked lists based on linguistic approaches. Our results contribute to the nascent field of typological interpretability of cross-lingual text representations.

Recently, domain-general recurrent neural networks, without explicit linguistic inductive biases, have been shown to successfully reproduce a range of human language behaviours, such as accurately predicting number agreement between nouns and verbs. We show that such networks will also learn number agreement within unnatural sentence structures, i.e. structures that are not found within any natural languages and which humans struggle to process. These results suggest that the models are learning from their input in a manner that is substantially different from human language acquisition, and we undertake an analysis of how the learned knowledge is stored in the weights of the network. We find that while the model has an effective understanding of singular versus plural for individual sentences, there is a lack of a unified concept of number agreement connecting these processes across the full range of inputs. Moreover, the weights handling natural and unnatural structures overlap substantially, in a way that underlines the non-human-like nature of the knowledge learned by the network.

The ability to change a person’s mind on a given issue depends both on the arguments they are presented with and on their underlying perspectives and biases on that issue. Predicting stance changes require characterizing both aspects and the interaction between them, especially in realistic settings in which stance changes are very rare. In this paper, we suggest a modular learning approach, which decomposes the task into multiple modules, focusing on different aspects of the interaction between users, their beliefs, and the arguments they are exposed to. Our experiments show that our modular approach archives significantly better results compared to the end-to-end approach using BERT over the same inputs.

Predicting users’ opinions in their response to social events has important real-world applications, many of which political and social impacts. Existing approaches derive a population’s opinion on a going event from large scores of user generated content. In certain scenarios, we may not be able to acquire such content and thus cannot infer an unbiased opinion on those emerging events. To address this problem, we propose to explore opinion on unseen articles based on one’s fingerprinting: the prior reading and commenting history. This work presents a focused study on modeling and leveraging fingerprinting techniques to predict a user’s future opinion. We introduce a recurrent neural network based model that integrates fingerprinting. We collect a large dataset that consists of event-comment pairs from six news websites. We evaluate the proposed model on this dataset. The results show substantial performance gains demonstrating the effectiveness of our approach.

Hoping for a large number of clicks and potentially high social shares, journalists of various news media outlets publish sensationalist headlines on social media. These headlines lure the readers to click on them and satisfy the curiosity gap in their mind. Low quality material pointed to by clickbaits leads to time wastage and annoyance for users. Even for enterprises publishing clickbaits, it hurts more than it helps as it erodes user trust, attracts wrong visitors, and produces negative signals for ranking algorithms. Hence, identifying and flagging clickbait titles is very essential. Previous work on clickbaits has majorly focused on binary classification of clickbait titles. However not all clickbaits are equally clickbaity. It is not only essential to identify a click-bait, but also to identify the intensity of the clickbait based on the strength of the clickbait. In this work, we model clickbait strength prediction as a regression problem. While previous methods have relied on traditional machine learning or vanilla recurrent neural networks, we rigorously investigate the use of transformers for clickbait strength prediction. On a benchmark dataset with ∼39K posts, our methods outperform all the existing methods in the Clickbait Challenge.

Active learning is able to significantly reduce the annotation cost for data-driven techniques. However, previous active learning approaches for natural language processing mainly depend on the entropy-based uncertainty criterion, and ignore the characteristics of natural language. In this paper, we propose a pre-trained language model based active learning approach for sentence matching. Differing from previous active learning, it can provide linguistic criteria from the pre-trained language model to measure instances and help select more effective instances for annotation. Experiments demonstrate our approach can achieve greater accuracy with fewer labeled training instances.

Incorporating lexicons into character-level Chinese NER by lattices is proven effective to exploitrich word boundary information. Previous work has extended RNNs to consume lattice inputsand achieved great success. However, due to the DAG structure and the inherently unidirectionalsequential nature, this method precludes batched computation and sufficient semantic interaction. In this paper, we propose PLTE, an extension of transformer encoder that is tailored for ChineseNER, which models all the characters and matched lexical words in parallel with batch process-ing. PLTE augments self-attention with positional relation representations to incorporate latticestructure. It also introduces a porous mechanism to augment localness modeling and maintainthe strength of capturing the rich long-term dependencies. Experimental results show that PLTEperforms up to 11.4 times faster than state-of-the-art methods while realizing better performance. We also demonstrate that using BERT representations further substantially boosts the performanceand brings out the best in PLTE.

We tackle the task of automatically generating a function name from source code. Existing generators face difficulties in generating low-frequency or out-of-vocabulary subwords. In this paper, we propose two strategies for copying low-frequency or out-of-vocabulary subwords in inputs. Our best performing model showed an improvement over the conventional method in terms of our modified F1 and accuracy on the Java-small and Java-large datasets.

Recurrent neural networks (RNNs) suffer from well-known limitations and complications which include slow inference and vanishing gradients when processing long sequences in text classification. Recent studies have attempted to accelerate RNNs via various ad hoc mechanisms to skip irrelevant words in the input. However, word skipping approaches proposed to date effectively stop at each or a given time step to decide whether or not a given input word should be skipped, breaking the coherence of input processing in RNNs. Furthermore, current methods cannot change skip rates during inference and are consequently unable to support different skip rates in demanding real-world conditions. To overcome these limitations, we propose Pointer- LSTM, a novel LSTM framework which relies on a pointer network to select important words for target prediction. The model maintains a coherent input process for the LSTM modules and makes it possible to change the skip rate during inference. Our evaluation on four public data sets demonstrates that Pointer-LSTM (a) is 1.1x∼3.5x faster than the standard LSTM architecture; (b) is more accurate than Leap-LSTM (the state-of-the-art LSTM skipping model) at high skip rates; and (c) reaches robust accuracy levels even when the skip rate is changed during inference.

Dependency context-based word embedding jointly learns the representations of word and dependency context, and has been proved effective in aspect term extraction. In this paper, we design the positional dependency-based word embedding (PoD) which considers both dependency context and positional context for aspect term extraction. Specifically, the positional context is modeled via relative position encoding. Besides, we enhance the dependency context by integrating more lexical information (e.g., POS tags) along dependency paths. Experiments on SemEval 2014/2015/2016 datasets show that our approach outperforms other embedding methods in aspect term extraction.

As the name implies, contextualized representations of language are typically motivated by their ability to encode context. Which aspects of context are captured by such representations? We introduce an approach to address this question using Representational Similarity Analysis (RSA). As case studies, we investigate the degree to which a verb embedding encodes the verb’s subject, a pronoun embedding encodes the pronoun’s antecedent, and a full-sentence representation encodes the sentence’s head word (as determined by a dependency parse). In all cases, we show that BERT’s contextualized embeddings reflect the linguistic dependency being studied, and that BERT encodes these dependencies to a greater degree than it encodes less linguistically-salient controls. These results demonstrate the ability of our approach to adjudicate between hypotheses about which aspects of context are encoded in representations of language.

A large body of research on gender-linked language has established foundations regarding cross-gender differences in lexical, emotional, and topical preferences, along with their sociological underpinnings. We compile a novel, large and diverse corpus of spontaneous linguistic productions annotated with speakers’ gender, and perform a first large-scale empirical study of distinctions in the usage of figurative language between male and female authors. Our analyses suggest that (1) idiomatic choices reflect gender-specific lexical and semantic preferences in general language, (2) men’s and women’s idiomatic usages express higher emotion than their literal language, with detectable, albeit more subtle, differences between male and female authors along the dimension of dominance compared to similar distinctions in their literal utterances, and (3) contextual analysis of idiomatic expressions reveals considerable differences, reflecting subtle divergences in usage environments, shaped by cross-gender communication styles and semantic biases.

Neural Machine Translation (NMT) has shown drastic improvement in its quality when translating clean input, such as text from the news domain. However, existing studies suggest that NMT still struggles with certain kinds of input with considerable noise, such as User-Generated Contents (UGC) on the Internet. To make better use of NMT for cross-cultural communication, one of the most promising directions is to develop a model that correctly handles these expressions. Though its importance has been recognized, it is still not clear as to what creates the great gap in performance between the translation of clean input and that of UGC. To answer the question, we present a new dataset, PheMT, for evaluating the robustness of MT systems against specific linguistic phenomena in Japanese-English translation. Our experiments with the created dataset revealed that not only our in-house models but even widely used off-the-shelf systems are greatly disturbed by the presence of certain phenomena.

The automatic feedback of school assignments is an important application of AI in education. In this work, we focus on the task of personalized multimodal feedback generation, which aims to generate personalized feedback for teachers to evaluate students’ assignments involving multimodal inputs such as images, audios, and texts. This task involves the representation and fusion of multimodal information and natural language generation, which presents the challenges from three aspects: (1) how to encode and integrate multimodal inputs; (2) how to generate feedback specific to each modality; and (3) how to fulfill personalized feedback generation. In this paper, we propose a novel Personalized Multimodal Feedback Generation Network (PMFGN) armed with a modality gate mechanism and a personalized bias mechanism to address these challenges. Extensive experiments on real-world K-12 education data show that our model significantly outperforms baselines by generating more accurate and diverse feedback. In addition, detailed ablation experiments are conducted to deepen our understanding of the proposed framework.

In the paper, we revisit the PP attachment problem which has been identified as one of the major sources for parser errors and discuss shortcomings of recent work. In particular, we show that using gold information for the extraction of attachment candidates as well as a missing comparison of the system’s output to the output of a full syntactic parser leads to an overly optimistic assessment of the results. We address these issues by presenting a realistic evaluation of the potential of different PP attachment systems, using fully predicted information as system input. We compare our results against the output of a strong neural parser and show that the full parsing approach is superior to modeling PP attachment disambiguation as a separate task.

Text-to-SQL is a task of translating utterances to SQL queries, and most existing neural approaches of text-to-SQL focus on the cross-domain context-independent generation task. We pay close attention to the cross-domain context-dependent text-to-SQL generation task, which requires a model to depend on the interaction history and current utterance to generate SQL query. In this paper, we present an encoder-decoder model called PG-GSQL based on the interaction-level encoder and with two effective innovations in decoder to solve cross-domain context-dependent text-to-SQL task. 1) To effectively capture historical information of SQL query and reuse the previous SQL query tokens, we use a hybrid pointer-generator network as decoder to copy tokens from the previous SQL query via pointer, the generator part is utilized to generate new tokens. 2) We propose a guide component to limit the prediction space of vocabulary for avoiding table-column dependency and foreign key dependency errors during decoding phase. In addition, we design a column-table linking mechanism to improve the prediction accuracy of tables. On the challenging cross-domain context-dependent text-to-SQL benchmark SParC, PG-GSQL achieves 34.0% question matching accuracy and 19.0% interaction matching accuracy on the dev set. With BERT augmentation, PG-GSQL obtains 53.1% question matching accuracy and 34.7% interaction matching accuracy on the dev set, outperforms the previous state-of-the-art model by 5.9% question matching accuracy and 5.2% interaction matching accuracy. Our code is publicly available.

Endowing a chatbot with a personality is essential to deliver more realistic conversations. Various persona-based dialogue models have been proposed to generate personalized and diverse responses by utilizing predefined persona information. However, generating personalized responses is still a challenging task since the leverage of predefined persona information is often insufficient. To alleviate this problem, we propose a novel Persona Enhanced Dual Alternating Learning Network (PEDNet) aiming at producing more personalized responses in various open-domain conversation scenarios. PEDNet consists of a Context-Dominate Network (CDNet) and a Persona-Dominate Network (PDNet), which are built upon a common encoder-decoder backbone. CDNet learns to select a proper persona as well as ensure the contextual relevance of the predicted response, while PDNet learns to enhance the utilization of persona information when generating the response by weakening the disturbance of specific content in the conversation context. CDNet and PDNet are trained alternately using a multi-task training approach to equip PEDNet with the both capabilities they have learned. Both automatic and human evaluations on a newly released dialogue dataset Persona-chat demonstrate that our method could deliver more personalized responses than baseline methods.

Dialog state tracking (DST) suffers from severe data sparsity. While many natural language processing (NLP) tasks benefit from transfer learning and multi-task learning, in dialog these methods are limited by the amount of available data and by the specificity of dialog applications. In this work, we successfully utilize non-dialog data from unrelated NLP tasks to train dialog state trackers. This opens the door to the abundance of unrelated NLP corpora to mitigate the data sparsity issue inherent to DST.

Language pairs with limited amounts of parallel data, also known as low-resource languages, remain a challenge for neural machine translation. While the Transformer model has achieved significant improvements for many language pairs and has become the de facto mainstream architecture, its capability under low-resource conditions has not been fully investigated yet. Our experiments on different subsets of the IWSLT14 training data show that the effectiveness of Transformer under low-resource conditions is highly dependent on the hyper-parameter settings. Our experiments show that using an optimized Transformer for low-resource conditions improves the translation quality up to 7.3 BLEU points compared to using the Transformer default settings.