A Context-Dependent Gated Module for Incorporating Symbolic Semantics into Event Coreference Resolution

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 3491–3499 June 6–11, 2021. ©2021 Association for Computational Linguistics 3491 A Context-Dependent Gated Module for Incorporating Symbolic Semantics into Event Coreference Resolution Tuan Lai, Heng Ji University of Illinois at Urbana-Champaign {tuanml2, hengji}@illinois.edu Trung Bui, Quan Hung Tran, Franck Dernoncourt, Walter Chang Adobe Research {bui, qtran, franck.dernoncourt, wachang}@adobe.com Abstract Event coreference resolution is an important research problem with many applications. De- spite the recent remarkable success of pre- trained language models, we argue that it is still highly beneficial to utilize symbolic fea- tures for the task. However, as the input for coreference resolution typically comes from upstream components in the information ex- traction pipeline, the automatically extracted symbolic features can be noisy and contain er- rors. Also, depending on the specific context, some features can be more informative than others. Motivated by these observations, we propose a novel context-dependent gated mod- ule to adaptively control the information flows from the input symbolic features. Combined with a simple noisy training method, our best models achieve state-of-the-art results on two datasets: ACE 2005 and KBP 2016.1 1 Introduction Within-document event coreference resolution is the task of clustering event mentions in a text that refer to the same real-world events (Lu and Ng, 2018). It is an important research problem, with many applications (Vanderwende et al., 2004; Ji and Grishman, 2011; Choubey et al., 2018). Since the trigger of an event mention is typically the word or phrase that most clearly describes the event, vir- tually all previous approaches employ features re- lated to event triggers in one form or another. To achieve better performance, many methods also need to use a variety of additional symbolic fea- tures such as event types, attributes, and arguments (Chen et al., 2009; Chen and Ji, 2009; Zhang et al., 2015; Sammons et al., 2015; Lu and Ng, 2016; Chen and Ng, 2016; Duncan et al., 2017). Previous neural methods (Nguyen et al., 2016; Choubey and Huang, 2017; Huang et al., 2019) also use non- contextual word embeddings such as word2vec 1The code is publicly available at https://github.com/ laituan245/eventcoref. ... we are seeing these soldiers {head out}ev1 ... ... these soldiers were set to {leave}ev2 in January ... ev1 (Movement:Transport): Modality = ASSERTED ev2 (Movement:Transport): Modality = OTHER Table 1: An example of using the modality attribute to improve event coreference resolution. (Mikolov et al., 2013) or GloVe (Pennington et al., 2014). With the recent remarkable success of language models such as BERT (Devlin et al., 2019) and SpanBERT (Joshi et al., 2020), one natural ques- tion is whether we can simply use these models for coreference resolution without relying on any additional features. We argue that it is still highly beneficial to utilize symbolic features, especially when they are clean and have complementary in- formation. Table 1 shows an example in the ACE 2005 dataset, where our baseline SpanBERT model incorrectly predicts the highlighted event …

Content truncated to 500 words