COLING 2022 Papers

In this paper, we propose to leverage the unique characteristics of dialogues sharing commonsense knowledge across participants, to resolve the difficulties in summarizing them. We present SICK, a framework that uses commonsense inferences as additional context. Compared to previous work that solely relies on the input dialogue, SICK uses an external knowledge model to generate a rich set of commonsense inferences and selects the most probable one with a similarity-based selection method. Built upon SICK, SICK++ utilizes commonsense as supervision, where the task of generating commonsense inferences is added upon summarizing the dialogue in a multi-task learning setting. Experimental results show that with injected commonsense knowledge, our framework generates more informative and consistent summaries than existing methods.

In the field of Natural Language Processing (NLP), extracting method entities from biomedical text has been a challenging task. Scientific research papers commonly consist of complex keywords and domain-specific terminologies, and new terminologies are continuously appearing. In this research, we find method terminologies in biomedical text using both rule-based and machine learning techniques. We first use linguistic features to extract method sentence candidates from a large corpus of biomedical text. Then, we construct a silver standard biomedical corpus composed of these sentences. With a rule-based method that makes use of the Stanza dependency parsing module, we label the method entities in these sentences. Using this silver standard corpus we train two machine learning algorithms to automatically extract method entities from biomedical text. Our results show that it is possible to develop machine learning models that can automatically extract method entities to a reasonable accuracy without the need for a gold standard dataset.

Linguists distinguish between novel and conventional metaphor, a distinction which the metaphor detection task in NLP does not take into account. Instead, metaphoricity is formulated as a property of a token in a sentence, regardless of metaphor type. In this paper, we investigate the limitations of treating conventional metaphors in this way, and advocate for an alternative which we name ‘metaphorical polysemy detection’ (MPD). In MPD, only conventional metaphoricity is treated, and it is formulated as a property of word senses in a lexicon. We develop the first MPD model, which learns to identify conventional metaphors in the English WordNet. To train it, we present a novel training procedure that combines metaphor detection with ‘word sense disambiguation’ (WSD). For evaluation, we manually annotate metaphor in two subsets of WordNet. Our model significantly outperforms a strong baseline based on a state-of-the-art metaphor detection model, attaining an ROC-AUC score of .78 (compared to .65) on one of the sets. Additionally, when paired with a WSD model, our approach outperforms a state-of-the-art metaphor detection model at identifying conventional metaphors in text (.659 F1 compared to .626).

In this paper we present MisNet, a novel model for word level metaphor detection. MisNet converts two linguistic rules, i.e., Metaphor Identification Procedure (MIP) and Selectional Preference Violation (SPV) into semantic matching tasks. MIP module computes the similarity between the contextual meaning and the basic meaning of a target word. SPV module perceives the incongruity between target words and their contexts. To better represent basic meanings, MisNet utilizes dictionary resources. Empirical results indicate that MisNet achieves competitive performance on several datasets.

Due to the lack of labeled data in many realistic scenarios, a number of few-shot learning methods for text classification have been proposed, among which the meta learning based ones have recently attracted much attention. Such methods usually consist of a learner as the classifier and a meta learner for specializing the learner to different tasks. For the learner, learning rate is crucial to its performance. However, existing methods treat it as a hyper parameter and adjust it manually, which is time-consuming and laborious. Intuitively, for different tasks and neural network layers, the learning rates should be different and self-adaptive. For the meta learner, it requires a good generalization ability so as to quickly adapt to new tasks. Motivated by these issues, we propose a novel meta learning framework, called MetaSLRCL, for few-shot text classification. Specifically, we present a novel meta learning mechanism to obtain different learning rates for different tasks and neural network layers so as to enable the learner to quickly adapt to new training data. Moreover, we propose a task-oriented curriculum learning mechanism to help the meta learner achieve a better generalization ability by learning from different tasks with increasing difficulties. Extensive experiments on three benchmark datasets demonstrate the effectiveness of MetaSLRCL.

Prompting method is regarded as one of the crucial progress for few-shot nature language processing. Recent research on prompting moves from discrete tokens based “hard prompts” to continuous “soft prompts”, which employ learnable vectors as pseudo prompt tokens and achieve better performance. Though showing promising prospects, these soft-prompting methods are observed to rely heavily on good initialization to take effect. Unfortunately, obtaining a perfect initialization for soft prompts requires understanding of inner language models working and elaborate design, which is no easy task and has to restart from scratch for each new task. To remedy this, we propose a generalized soft prompting method called MetaPrompting, which adopts the well-recognized model-agnostic meta-learning algorithm to automatically find better prompt initialization that facilitates fast adaptation to new prompting tasks. Extensive experiments show MetaPrompting tackles soft prompt initialization problem and brings significant improvement on three different datasets (over 7 points improvement in accuracy for 1-shot setting), achieving new state-of-the-art performance.

Complex question generation over knowledge bases (KB) aims to generate natural language questions involving multiple KB relations or functional constraints. Existing methods train one encoder-decoder-based model to fit all questions. However, such a one-size-fits-all strategy may not perform well since complex questions exhibit an uneven distribution in many dimensions, such as question types, involved KB relations, and query structures, resulting in insufficient learning for long-tailed samples under different dimensions. To address this problem, we propose a meta-learning framework for complex question generation. The meta-trained generator can acquire universal and transferable meta-knowledge and quickly adapt to long-tailed samples through a few most related training samples. To retrieve similar samples for each input query, we design a self-supervised graph retriever to learn distributed representations for samples, and contrastive learning is leveraged to improve the learned representations. We conduct experiments on both WebQuestionsSP and ComplexWebQuestion, and results on long-tailed samples of different dimensions have been significantly improved, which demonstrates the effectiveness of the proposed framework.

Cross-domain sentiment analysis aims to predict the sentiment of texts in the target domain using the model trained on the source domain to cope with the scarcity of labeled data. Previous studies are mostly cross-entropy-based methods for the task, which suffer from instability and poor generalization. In this paper, we explore contrastive learning on the cross-domain sentiment analysis task. We propose a modified contrastive objective with in-batch negative samples so that the sentence representations from the same class can be pushed close while those from the different classes become further apart in the latent space. Experiments on two widely used datasets show that our model can achieve state-of-the-art performance in both cross-domain and multi-domain sentiment analysis tasks. Meanwhile, visualizations demonstrate the effectiveness of transferring knowledge learned in the source domain to the target domain and the adversarial test verifies the robustness of our model.

Current medical question answering systems have difficulty processing long, detailed and informally worded questions submitted by patients, called Consumer Health Questions (CHQs). To address this issue, we introduce a medical question understanding and answering system with knowledge grounding and semantic self-supervision. Our system is a pipeline that first summarizes a long, medical, user-written question, using a supervised summarization loss. Then, our system performs a two-step retrieval to return answers. The system first matches the summarized user question with an FAQ from a trusted medical knowledge base, and then retrieves a fixed number of relevant sentences from the corresponding answer document. In the absence of labels for question matching or answer relevance, we design 3 novel, self-supervised and semantically-guided losses. We evaluate our model against two strong retrieval-based question answering baselines. Evaluators ask their own questions and rate the answers retrieved by our baselines and own system according to their relevance. They find that our system retrieves more relevant answers, while achieving speeds 20 times faster. Our self-supervised losses also help the summarizer achieve higher scores in ROUGE, as well as in human evaluation metrics.

Relation extraction in the biomedical domain is challenging due to the lack of labeled data and high annotation costs, needing domain experts. Distant supervision is commonly used to tackle the scarcity of annotated data by automatically pairing knowledge graph relationships with raw texts. Such a pipeline is prone to noise and has added challenges to scale for covering a large number of biomedical concepts. We investigated existing broad-coverage distantly supervised biomedical relation extraction benchmarks and found a significant overlap between training and test relationships ranging from 26% to 86%. Furthermore, we noticed several inconsistencies in the data construction process of these benchmarks, and where there is no train-test leakage, the focus is on interactions between narrower entity types. This work presents a more accurate benchmark MedDistant19 for broad-coverage distantly supervised biomedical relation extraction that addresses these shortcomings and is obtained by aligning the MEDLINE abstracts with the widely used SNOMED Clinical Terms knowledge base. Lacking thorough evaluation with domain-specific language models, we also conduct experiments validating general domain relation extraction findings to biomedical relation extraction.

The quality of Natural Language Processing (NLP) models is typically measured by the accuracy or error rate of a predefined test set. Because the evaluation and optimization of these measures are narrowed down to a specific domain like news and cannot be generalized to other domains like Twitter, we often observe that a system reported with human parity results generates surprising errors in real-life use scenarios. We address this weakness with a new approach that uses an NLP quality measure based on robustness. Unlike previous work that has defined robustness using Minimax to bound worst cases, we measure robustness based on the consistency of cross-domain accuracy and introduce the coefficient of variation and (epsilon, gamma)-Robustness. Our measures demonstrate higher agreements with human evaluation than accuracy scores like BLEU on ranking Machine Translation (MT) systems. Our experiments of sentiment analysis and MT tasks show that incorporating our robustness measures into learning objectives significantly enhances the final NLP prediction accuracy over various domains, such as biomedical and social media.

Morphological systems across languages vary when it comes to the relation between form and meaning. In some languages, a single meaning feature corresponds to a single morpheme, whereas in other languages, multiple meaning features are bundled together into one morpheme. The two types of languages have been called agglutinative and fusional, respectively, but this distinction does not capture the graded nature of the phenomenon. We provide a mathematically precise way of characterizing morphological systems using partial information decomposition, a framework for decomposing mutual information into three components: unique, redundant, and synergistic information. We show that highly fusional languages are characterized by high levels of synergy.

Text classifiers are applied at scale in the form of one-size-fits-all solutions. Nevertheless, many studies show that classifiers are biased regarding different languages and dialects. When measuring and discovering these biases, some gaps present themselves and should be addressed. First, “Does language, dialect, and topical content vary across geographical regions?” and secondly “If there are differences across the regions, do they impact model performance?”. We introduce a novel dataset called GeoOLID with more than 14 thousand examples across 15 geographically and demographically diverse cities to address these questions. We perform a comprehensive analysis of geographical-related content and their impact on performance disparities of offensive language detection models. Overall, we find that current models do not generalize across locations. Likewise, we show that while offensive language models produce false positives on African American English, model performance is not correlated with each city’s minority population proportions. Warning: This paper contains offensive language.

We present a subword regularization method for WordPiece, which uses a maximum matching algorithm for tokenization. The proposed method, MaxMatch-Dropout, randomly drops words in a search using the maximum matching algorithm. It realizes finetuning with subword regularization for popular pretrained language models such as BERT-base. The experimental results demonstrate that MaxMatch-Dropout improves the performance of text classification and machine translation tasks as well as other subword regularization methods. Moreover, we provide a comparative analysis of subword regularization methods: subword regularization with SentencePiece (Unigram), BPE-Dropout, and MaxMatch-Dropout.

Large pre-trained language models (PLMs) have demonstrated superior performance in industrial applications. Recent studies have explored parameter-efficient PLM tuning, which only updates a small amount of task-specific parameters while achieving both high efficiency and comparable performance against standard fine-tuning. However, all these methods ignore the inefficiency problem caused by the task-specific output layers, which is inflexible for us to re-use PLMs and introduces non-negligible parameters. In this work, we focus on the text classification task and propose plugin-tuning, a framework that further improves the efficiency of existing parameter-efficient methods with a unified classifier. Specifically, we re-formulate both token and sentence classification tasks into a unified language modeling task, and map label spaces of different tasks into the same vocabulary space. In this way, we can directly re-use the language modeling heads of PLMs, avoiding introducing extra parameters for different tasks. We conduct experiments on six classification benchmarks. The experimental results show that plugin-tuning can achieve comparable performance against fine-tuned PLMs, while further saving around 50% parameters on top of other parameter-efficient methods.

Two of the most fundamental issues in Natural Language Understanding (NLU) at present are: (a) how it can established whether deep learning-based models score highly on NLU benchmarks for the ”right” reasons; and (b) what those reasons would even be. We investigate the behavior of reading comprehension models with respect to two linguistic ”skills”: coreference resolution and comparison. We propose a definition for the reasoning steps expected from a system that would be ”reading slowly”, and compare that with the behavior of five models of the BERT family of various sizes, observed through saliency scores and counterfactual explanations. We find that for comparison (but not coreference) the systems based on larger encoders are more likely to rely on the ”right” information, but even they struggle with generalization, suggesting that they still learn specific lexical patterns rather than the general principles of comparison.

In contrast to traditional exhaustive search, selective search first clusters documents into several groups before all the documents are searched exhaustively by a query, to limit the search executed within one group or only a few groups. Selective search is designed to reduce the latency and computation in modern large-scale search systems. In this study, we propose MICO, a Mutual Information CO-training framework for selective search with minimal supervision using the search logs. After training, MICO does not only cluster the documents, but also routes unseen queries to the relevant clusters for efficient retrieval. In our empirical experiments, MICO significantly improves the performance on multiple metrics of selective search and outperforms a number of existing competitive baselines.

Event Causality Identification (ECI) is the task of detecting causal relations between events mentioned in the text. Although this task has been extensively studied for English materials, it is under-explored for many other languages. A major reason for this issue is the lack of multilingual datasets that provide consistent annotations for event causality relations in multiple non-English languages. To address this issue, we introduce a new multilingual dataset for ECI, called MECI. The dataset employs consistent annotation guidelines for five typologically different languages, i.e., English, Danish, Spanish, Turkish, and Urdu. Our dataset thus enable a new research direction on cross-lingual transfer learning for ECI. Our extensive experiments demonstrate high quality for MECI that can provide ample research challenges and directions for future research. We will publicly release MECI to promote research on multilingual ECI.

This paper introduces a generative system for in-battle real-time commentary in mobile MOBA games. Event commentary is important for battles in MOBA games, which is applicable to a wide range of scenarios like live streaming, e-sports commentary and combat information analysis. The system takes real-time match statistics and events as input, and an effective transform method is designed to convert match statistics and utterances into consistent encoding space. This paper presents the general framework and implementation details of the proposed system, and provides experimental results on large-scale real-world match data.