Utilizing Relative Event Time to Enhance Event-Event Temporal Relation Extraction

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 10431–10437 November 7–11, 2021. c⃝2021 Association for Computational Linguistics 10431 Utilizing Relative Event Time to Enhance Event-Event Temporal Relation Extraction Haoyang Wen and Heng Ji Department of Computer Science University of Illinois Urbana-Champaign {wen17, hengji}@illinois.edu Abstract Event time is one of the most important fea- tures for event-event temporal relation extrac- tion. However, explicit event time informa- tion in text is sparse. For example, only about 20% of event mentions in TimeBank- Dense have event-time links. In this paper, we propose a joint model for event-event tem- poral relation classification and an auxiliary task, relative event time prediction, which pre- dicts the event time as real numbers. We adopt a Stack-Propagation framework to in- corporate predicted relative event time for temporal relation classification and keep the differentiability. Our experiments on MA- TRES dataset show that our model can signif- icantly improve the RoBERTa-based baseline and achieve state-of-the-art performance. 1 1 Introduction Event temporal ordering is an important task to understand the evolution of events. Event-event temporal relation extraction aims to automatically extract the temporal order given a pair of events and further build a temporal graph. Figure 1 illustrates an example sentence and its temporal graph. There are three events in the sentence, said, identified and run. The temporal relation between said and identified is AFTER, and the temporal relations between said and run and between identified and run are BEFORE. Neural network based methods have achieved promising improvement for temporal relation ex- traction (Meng et al., 2017; Meng and Rumshisky, 2018; Cheng et al., 2020; Ballesteros et al., 2020; Wen et al., 2021a). They mostly consider the task as pairwise classification. There are also efforts fo- cusing on the global structures, including Markov logical networks and Integer Linear Programming 1 The resource for this paper is available at https: //github.com/wenhycs/EMNLP2021-Utilizing -Relative-Event-Time-to-Enhance-Event-Ev ent-Temporal-Relation-Extraction After Before Before said run identified Microsoft (e1, said) it has (e2, identified) three compa- nies for the China program to (e3, run) through June . Figure 1: An example sentence and their temporal rela- tions. In this example, there three different events, and the final extracted graph shows the pair-wise temporal relation extraction results. (ILP) based methods (Bramsen et al., 2006; Cham- bers and Jurafsky, 2008; Yoshikawa et al., 2009; Do et al., 2012; Ning et al., 2017, 2018a, 2019; Han et al., 2019). Though achieving great success, event time, an important feature, is often overlooked by previous work. Conceptually, if we know the ex- act time information for all events, their temporal relations can be naturally derived. For example, if we know that event A happened on Monday while event B happened on Tuesday in the same week, it is obvious that A happened BEFORE B. How- ever, explicit time arguments can be rarely found in text, especially in news articles (Wen et al., 2021b). Leeuwenberg and Moens (2018) propose to pre- dict the relative timeline and directly compare the relative timestamps …

Content truncated to 500 words