Fine-Grained Event Trigger Detection

Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics, pages 2745–2752 April 19 - 23, 2021. ©2021 Association for Computational Linguistics 2745 Fine-Grained Event Trigger Detection Duong Minh Le VinAI Research, Vietnam v.duonglm1@vinai.io Thien Huu Nguyen∗ Department of Computer and Information Science University of Oregon, Eugene, OR 97403, USA thien@cs.uoregon.edu Abstract Most of the previous work on Event Detection (ED) has only considered the datasets with a small number of event types (i.e., up to 38 types). In this work, we present the first study on fine-grained ED (FED) where the evalua- tion dataset involves much more fine-grained event types (i.e., 449 types). We propose a novel method to transform the Semcor dataset for Word Sense Disambiguation into a large and high-quality dataset for FED. Extensive evaluation of the current ED methods is con- ducted to demonstrate the challenges of the generated datasets for FED, calling for more research effort in this area. 1 Introduction Understanding events in text is an important aspect of Natural Language Processing (NLP). Toward this end, Event Detection (ED), a task of Informa- tion Extraction (IE), aims to identify event triggers in sentences and classify them into some prede- fined types of interest. Event triggers represent the most important words (usually single verbs or nom- inalizations) in the sentences that evoke the events. The current state-of-the-art methods for ED feature the deep learning models where many new network architectures are introduced in the last couple of years (Nguyen and Grishman, 2015; Chen et al., 2015; Liu et al., 2017, 2019a; Lai et al., 2020b). Among others, the rapid development of the deep learning models for ED can be partly at- tributed to the availability of the large datasets to evaluate the models (e.g., the ACE 2005 and TAC KBP 2015 datasets (Walker et al., 2006; Mitamura et al., 2015)). Unfortunately, a major issue in these existing datasets for ED is that they tend to only focus on a limited set of event types. For example, the popular ACE 2005 dataset is only annotated ∗Corresponding author. for 33 event subtypes (e.g., Attack, Start-Position, Elect) while the number of events in the TAC KBP dataset (Mitamura et al., 2015) is 38. On the one hand, the limited numbers of types are unable to cover a wide range of possible events in practice (Araki and Mitamura, 2018). On the other hand, the small label sets often amount to the coarse-grained event types in the existing datasets that cannot cap- ture the slightly different nuances (i.e., fine-grained distinction) of the events. For instance, both the words “quit” and “fired” in the two sentences “He decided to quit the job.” and “He was fired due to a policy violation.” (respectively) would be consid- ered as the trigger words of the same event type of End-Position in the ACE 2005 dataset. However, the nuances in these two events are quite different (i.e., in term of the willingness of the job termi- nation) and the ability …

Content truncated to 500 words