MRN: A Locally and Globally Mention-Based Reasoning Network for Document-Level Relation Extraction

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 1359–1370 August 1–6, 2021. ©2021 Association for Computational Linguistics 1359 MRN: A Locally and Globally Mention-Based Reasoning Network for Document-Level Relation Extraction Jingye Li1, Kang Xu1, Fei Li1∗, Hao Fei1, Yafeng Ren2, Donghong Ji1 1Key Laboratory of Aerospace Information Security and Trusted Computing, Ministry of Education, School of Cyber Science and Engineering, Wuhan University, China 2Laboratory of Language and Artificial Intelligence, Guangdong University of Foreign Studies, China {theodorelee,xukang0531,lifei csnlp,hao.fei,renyafeng,dhji}@whu.edu.cn Abstract Document-level relation extraction aims to de- tect the relations within one document, which is challenging since it requires complex rea- soning using mentions, entities, local and global contexts. Few previous studies have dis- tinguished local and global reasoning explic- itly, which may be problematic because they play different roles in intra- and inter-sentence relations. Moreover, the interactions between local and global contexts should be considered since they could help relation reasoning based on our observation. In this paper, we pro- pose a novel mention-based reasoning (MRN) module based on explicitly and collaboratively local and global reasoning. Based on MRN, we design a co-predictor module to predict entity relations based on local and global en- tity and relation representations jointly. We evaluate our MRN model on three widely- used benchmark datasets, namely DocRED, CDR, and GDA. Experimental results show that our model outperforms previous state-of- the-art models by a large margin. 1 Introduction Relation extraction (RE), identifying the semantic relations among target entities in the text, has long been a fundamental task in the natural language pro- cessing (NLP) community (Zeng et al., 2014; Xu et al., 2015). Prior efforts largely focus on sentence- level RE (Lin et al., 2016; Zhang et al., 2018). How- ever, recent studies reveal that a large number of relations can actually be expressed through multi- ple sentences, which necessitates document-level RE (Yao et al., 2019). Compared with sentence- level RE, the entities for document-level relations may be mentioned in multiple sentences across a document. Therefore, document-level RE requires capturing the complex interactions between all enti- ∗Corresponding author. P570 for the position of a Chicago municipal court judge in 1939. object subject: intra-sentence relation object subject: inter-sentence relation P17: country P27: country of citizenship P102: member of political party P570: date of death P27 [S1] Edward Rowan Finnegan (June 5, 1905 – February 2, 1971) was a U.S. Representative from Illinois from 1961 to 1964. [S5] Finnegan unsuccessfully sought the Democratic nomination ... P17 [S12] He continued to serve until his death in 1971. ... P102 ... Figure 1: An example of document-level RE from the DocRED dataset. We use the same color to denote the mentions of the same entity. ties in the entire document (Nan et al., 2020; Zeng et al., 2020; Wang et al., 2020). It is well known that local and global contexts are two key performance enhancers for the task. Intuitively, the former can benefit the identification of nearer (e.g., intra-sentence) relations, while the latter is more useful for distant (e.g., …

Content truncated to 500 words