LearnDA: Learnable Knowledge-Guided Data Augmentation for Event Causality Identification

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pages 3558–3571 August 1–6, 2021. ©2021 Association for Computational Linguistics 3558 LearnDA: Learnable Knowledge-Guided Data Augmentation for Event Causality Identification Xinyu Zuo1,2, Pengfei Cao1,2, Yubo Chen1,2, Kang Liu1,2, Jun Zhao1,2, Weihua Peng3 and Yuguang Chen3 1National Laboratory of Pattern Recognition, Institute of Automation, CAS, Beijing, China 2School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China 3Beijing Baidu Netcom Science Technology Co., Ltd {xinyu.zuo,pengfei.cao,yubo.chen,kliu,jzhao}@nlpr.ia.ac.cn {pengweihua,chenyuguang}@baidu.com Abstract Modern models for event causality identifica- tion (ECI) are mainly based on supervised learning, which are prone to the data lack- ing problem. Unfortunately, the existing NLP- related augmentation methods cannot directly produce available data required for this task. To solve the data lacking problem, we intro- duce a new approach to augment training data for event causality identification, by iteratively generating new examples and classifying event causality in a dual learning framework. On the one hand, our approach is knowledge guided, which can leverage existing knowledge bases to generate well-formed new sentences. On the other hand, our approach employs a dual mechanism, which is a learnable augmenta- tion framework, and can interactively adjust the generation process to generate task-related sentences. Experimental results on two bench- marks EventStoryLine and Causal-TimeBank show that 1) our method can augment suit- able task-related training data for ECI; 2) our method outperforms previous methods on EventStoryLine and Causal-TimeBank (+2.5 and +2.1 points on F1 value respectively). 1 Introduction Event causality identification (ECI) aims to iden- tify causal relations between events in texts, which can provide crucial clues for NLP tasks, such as logical reasoning and question answering (Girju, 2003; Oh et al., 2013, 2017). This task is usually modeled as a classification problem, i.e. determin- ing whether there is a causal relation between two events in a sentence. For example in Figure 1, an ECI system should identify two causal relations in two sentences: (1) attack cause −→ killed in S1; (2) statement cause −→ protests in S2. Most existing methods for ECI heavily rely on annotated training data (Mirza and Tonelli, 2016; Kimani Gray, a young man who likes football, was killed in a police attack shortly after a tight match. In the week following the fatal violence, several protests have erupted because of the official statement. S1: S2: Kimani Gray, a young man who likes football, was killed in a police attack shortly after a tight match. EDA deletion S3: Figure 1: S1 and S2 are causal sentences that contain causal events. S3 is produced by EDA based on S1. The dotted line indicates the causal relation. Riaz and Girju, 2014b; Hashimoto et al., 2014; Hu and Walker, 2017; Gao et al., 2019). However, ex- isting datasets are relatively small, which impede the training of the high-performance event causality reasoning model. According to our statistics, the largest widely used dataset EventStoryLine Corpus (Caselli and Vossen, 2017) only contains 258 docu- ments, 4316 …

Content truncated to 500 words