MLBiNet: A Cross-Sentence Collective Event Detection Network

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pages 4829–4839 August 1–6, 2021. ©2021 Association for Computational Linguistics 4829 MLBiNet: A Cross-Sentence Collective Event Detection Network Dongfang Lou1,2 ∗, Zhilin Liao 1,2 ∗, Shumin Deng1,2, Ningyu Zhang1,2 †, Huajun Chen1,2 † 1 Zhejiang University & AZFT Joint Lab for Knowledge Engine 2 Hangzhou Innovation Center, Zhejiang University loudongfang2015@163.com, zhilinliao@yeah.net {231sm,zhangningyu,huajunsir}@zju.edu.cn Abstract We consider the problem of collectively de- tecting multiple events, particularly in cross- sentence settings. The key to dealing with the problem is to encode semantic information and model event inter-dependency at a document- level. In this paper, we reformulate it as a Seq2Seq task and propose a Multi-Layer Bidirectional Network (MLBiNet) to capture the document-level association of events and semantic information simultaneously. Specifi- cally, a bidirectional decoder is firstly devised to model event inter-dependency within a sen- tence when decoding the event tag vector se- quence. Secondly, an information aggregation module is employed to aggregate sentence- level semantic and event tag information. Fi- nally, we stack multiple bidirectional decoders and feed cross-sentence information, forming a multi-layer bidirectional tagging architecture to iteratively propagate information across sen- tences. We show that our approach provides significant improvement in performance com- pared to the current state-of-the-art results1. 1 Introduction Event detection (ED) is a crucial sub-task of event extraction, which aims to identify and classify event triggers. For instance, the document shown in Table 1, which contains six sentences {s1, . . . , s6}, the ED system is required to identify four events: an Injure event triggered by “injuries”, two Attack events triggered by “firing” and “fight”, and a Die event triggered by “death”. Detecting event triggers from natural language text is a challenge task because of the following problems: a). Sentence-level contextual repre- sentation and document-level information ag- gregation (Chen et al., 2018; Zhao et al., 2018; ∗ Equal contribution and shared co-first authorship. † Corresponding author. 1The code is available in https://github.com/ zjunlp/DocED. s1: what a brave young woman s2: did you hear about the injuries[Injure] she sustained s3: did you hear about the firing[Attack] she did s4: she was going to fight[Attack] to the death[Die] s5: she was captured but she was one tough cookie s6: god bless here Table 1: An example document in ACE 2005 corpus with cross-sentence semantic enhancement and event inter-dependency. Specifically, semantic information of s2 provides latent information to enhance s3, and Attack event in s4 also contributes to s3. Shen et al., 2020). In ACE 2005 corpus, the argu- ments of a single event instance may be scattered in multiple sentences (Zheng et al., 2019; Ebner et al., 2019), which indicates that document-level infor- mation aggregation is critical for ED task. What’s more, a word in different contexts would express different meanings and trigger different events. For example, in Table 1, “firing” in s3 means the ac- tion of firing guns (Attack event) or forcing some- body to …

Content truncated to 500 words