Better Highlighting: Creating Sub-Sentence Summary Highlights

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 684–695, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 684 Connecting the Dots: Event Graph Schema Induction with Path Language Modeling Manling Li1, Qi Zeng1, Ying Lin1, Kyunghyun Cho2, Heng Ji1, Jonathan May3, Nathanael Chambers4, Clare Voss5 1University of Illinois at Urbana-Champaign 2New York University 3University of Southern California 4US Naval Academy 5US Army Research Laboratory {manling2,qizeng2,yinglin8,hengji}@illinois.edu, kyunghyun.cho@nyu.edu, jonmay@isi.edu, nchamber@usna.edu, clare.r.voss.civ@mail.mil Abstract Event schemas can guide our understanding and ability to make predictions with respect to what might happen next. We propose a new Event Graph Schema, where two event types are connected through multiple paths involving entities that fill important roles in a coherent story. We then introduce Path Language Model, an auto-regressive language model trained on event-event paths, and se- lect salient and coherent paths to probabilisti- cally construct these graph schemas. We de- sign two evaluation metrics, instance cover- age and instance coherence, to evaluate the quality of graph schema induction, by check- ing when coherent event instances are covered by the schema graph. Intrinsic evaluations show that our approach is highly effective at inducing salient and coherent schemas. Ex- trinsic evaluations show the induced schema repository provides significant improvement to downstream end-to-end Information Extrac- tion over a state-of-the-art joint neural extrac- tion model, when used as additional global fea- tures to unfold instance graphs.1 1 Introduction Existing approaches to automated event extraction retain the overly simplistic assumption that events are atomic occurrences. Understanding events re- quires knowledge in the form of a repository of ab- stracted event schemas (complex event templates). Scripts (Schank and Abelson, 1977) encode fre- quently recurring event sequences, where events are ordered by temporal relation (Chambers and Ju- rafsky, 2009), causal relation (Mostafazadeh et al., 2016b), or narrative order (Jans et al., 2012). Event schemas have become increasingly important for natural language understanding tasks such as story 1Our code and data are publicly available for research pur- pose at http://blender.cs.illinois.edu/software/ pathlm. ending prediction (Mostafazadeh et al., 2016a) and reading comprehension (Koˇcisk´y et al., 2018; Os- termann et al., 2019). Previous schema induction methods mostly ig- nore uncertainty, re-occurring events and multiple hypotheses, with limited attention to capture com- plex relations among events, other than temporal or causal relations. Temporal relations exist between almost all events, even those that are not semanti- cally related; while research in identifying causal relations has been hobbled by low inter-annotator agreement (Hong et al., 2016). In this paper, we hypothesize that two events are connected when their entity arguments are co- referential or semantically related. For example, in Figure 1, (a) and (b) refer to very different event instances, but they both illustrate a typical scenario where a group of people moved from one place to another and then attacked the destination. From many such event instance pairs, we can induce multiple paths connecting a movement event to a related attack event: the person being moved became the attacker, and the weapon or vehicle …

Content truncated to 500 words