DEGREE: A Data-Efficient Generation-Based Event Extraction Model

Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 1890 - 1908 July 10-15, 2022 ©2022 Association for Computational Linguistics DEGREE: A Data-Efficient Generation-Based Event Extraction Model I-Hung Hsu∗† Kuan-Hao Huang∗‡ Elizabeth Boschee† Scott Miller† Premkumar Natarajan† Kai-Wei Chang‡ Nanyun Peng†‡ †Information Science Institute, University of Southern California ‡Computer Science Department, University of California, Los Angeles {ihunghsu, boschee, smiller, pnataraj}@isi.edu {khhuang, kwchang, violetpeng}@cs.ucla.edu Abstract Event extraction requires high-quality expert human annotations, which are usually expen- sive. Therefore, learning a data-efficient event extraction model that can be trained with only a few labeled examples has become a crucial chal- lenge. In this paper, we focus on low-resource end-to-end event extraction and propose DE- GREE, a data-efficient model that formulates event extraction as a conditional generation problem. Given a passage and a manually de- signed prompt, DEGREE learns to summarize the events mentioned in the passage into a nat- ural sentence that follows a predefined pattern. The final event predictions are then extracted from the generated sentence with a determinis- tic algorithm. DEGREE has three advantages to learn well with less training data. First, our de- signed prompts provide semantic guidance for DEGREE to leverage label semantics and thus better capture the event arguments. Moreover, DEGREE is capable of using additional weakly- supervised information, such as the description of events encoded in the prompts. Finally, DE- GREE learns triggers and arguments jointly in an end-to-end manner, which encourages the model to better utilize the shared knowledge and dependencies among them. Our experimen- tal results demonstrate the strong performance of DEGREE for low-resource event extraction. 1 Introduction Event extraction (EE) aims to extract events, each of which consists of a trigger and several partici- pants (arguments) with their specific roles, from a given passage. For example, in Figure 1, a Jus- tice:Execute event is triggered by the word “execu- tion” and this event contains three argument roles, including an Agent (Indonesia) who carries out the execution, a Person who is executed (convicts), and a Place where the event occurs (not mentioned in the passage). Previous work (Yang et al., 2019a; ∗The authors contribute equally. Passage: Indonesia will delay the execution of six convicts including an Indian on death row after five of them appealed to the Supreme Court for a second review. Justice:Execute Justice:Appeal Person Plaintiff Agent Adjudicator Place Justice:Execute Agent : Indonesia Person : convicts Place : None Justice:Appeal Plaintiff : five Prosecutor : None Adjudicator : Supreme Court Place : Indonesia Figure 1: Two examples of events (Justice:Execute and Justice:Appeal) extracted from the given passage. Fincke et al., 2021) usually divides EE into two sub- tasks: (1) event detection, which identifies event triggers and their types, and (2) event argument extraction, which extracts the arguments and their roles for given event triggers. EE has been shown to benefit a wide range of applications, e.g., building knowledge graphs (Zhang et al., 2020), question answering (Berant et al., 2014; Han et al., …

Content truncated to 500 words