Resource-Enhanced Neural Model for Event Argument Extraction

Findings of the Association for Computational Linguistics: EMNLP 2020, pages 3554–3559 November 16 - 20, 2020. c⃝2020 Association for Computational Linguistics 3554 Resource-Enhanced Neural Model for Event Argument Extraction Jie Ma∗ Shuai Wang∗ Rishita Anubhai Miguel Ballesteros Yaser Al-Onaizan Amazon AI {jieman, wshui, ranubhai, ballemig, onaizan}@amazon.com Abstract Event argument extraction (EAE) aims to iden- tify the arguments of an event and classify the roles that those arguments play. Despite great efforts made in prior work, there remain many challenges: (1) Data scarcity. (2) Cap- turing the long-range dependency, specifically, the connection between an event trigger and a distant event argument. (3) Integrating event trigger information into candidate argument representation. For (1), we explore using un- labeled data in different ways. For (2), we propose to use a syntax-attending Transformer that can utilize dependency parses to guide the attention mechanism. For (3), we propose a trigger-aware sequence encoder with several types of trigger-dependent sequence represen- tations. We also support argument extraction either from text annotated with gold entities or from plain text. Experiments on the English ACE2005 benchmark show that our approach achieves a new state-of-the-art. 1 Introduction Event argument extraction (EAE) aims to identify the entities that serve as arguments of an event and to classify the specific roles they play. As in Fig. 1, “two soldiers” and “yesterday” are ar- guments, where the event triggers are “attacked” (with event type being ATTACK1) and “injured” (event type INJURY). For the trigger “attacked”, “two soldiers” plays the argument role Target while “yesterday” plays the argument role Attack Time. For the event trigger “injured”, “two soldiers” and “yesterday” play the role Victim and INJURY Time, respectively. There has been significant work on event extraction (EE) (Liao and Grishman, 2010; Hong et al., 2011; Li et al., 2013), but the EAE ∗Indicates Equal Contribution. 1Following ACE https://www.ldc.upenn.edu/ collaborations/past-projects/ace Figure 1: Event examples: Green indicates triggers with their types. Yellow indicates arguments. An arrow indicates the role played by an argument in a trigger. task remains a challenge and has become the bot- tleneck for improving the overall performance of EE (Wang et al., 2019a).2 Supervised data for EAE is expensive and hence scarce. One possible solution is to use other avail- able resources like unlabeled data. For that, (1) We use BERT (Devlin et al., 2018) as our model en- coder which leverages a much larger unannotated corpus where semantic information is captured. Un- like Yang et al. (2019) who added a final/prediction layer to BERT for argument extraction, we use BERT as token embedder and build a sequence of EAE task-specific components (Sec. 2). (2) We use (unlabeled) in-domain data to adapt the BERT model parameters in a subsequent pretraining step as in (Gururangan et al., 2020). This makes the en- coder domain-aware. (3) We perform self-training to construct auto-labeled data (silver data). A crucial aspect for EAE is to integrate event trigger information into the learned representations. This is important because arguments are depen- dent on triggers, i.e., the same argument …

Content truncated to 500 words