Learning Event Graph Knowledge for Abductive Reasoning

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pages 5181–5190 August 1–6, 2021. ©2021 Association for Computational Linguistics 5181 Learning Event Graph Knowledge for Abductive Reasoning Li Du, Xiao Ding∗, Ting Liu, and Bing Qin Research Center for Social Computing and Information Retrieval Harbin Institute of Technology, China {ldu, xding, tliu, qinb}@ir.hit.edu.cn Abstract Abductive reasoning aims at inferring the most plausible explanation for observed events, which would play critical roles in various NLP applications, such as reading comprehension and question answering. To facilitate this task, a narrative text based abductive reasoning task αNLI is proposed, together with explorations about building reasoning framework using pre- trained language models. However, abundant event commonsense knowledge is not well ex- ploited for this task. To fill this gap, we pro- pose a variational autoencoder based model ege-RoBERTa, which employs a latent variable to capture the necessary commonsense knowl- edge from event graph for guiding the abduc- tive reasoning task. Experimental results show that through learning the external event graph knowledge, our approach outperforms the base- line methods on the αNLI task. 1 Introduction Abductive reasoning aims at seeking for the best ex- planations for incomplete observations (Bhagavat- ula et al., 2019). For example, given observations Forgot to close window when leaving home and The room was in a mess, human beings can generate a reasonable hypothesis for explaining the obser- vations, such as A thief entered the room based on commonsense knowledge in their mind. However, due to the lack of commonsense knowledge and effective reasoning mechanism, this is still a chal- lenging problem for today’s cognitive intelligent systems (Charniak and Shimony, 1990; Oh et al., 2013; Kruengkrai et al., 2017). Most previous works focus on conducting ab- ductive reasoning based on formal logic (Eshghi et al., 1988; Levesque, 1989; Ng et al., 1990; Paul, 1993). However, the rigidity of formal logic lim- its the application of abductive reasoning in NLP ∗Corresponding author Figure 1: (a) An example of abductive reasoning. (b) Additional commonsense knowledge (such as event I1 and I2) is necessary for inferring the correct hypothe- sis. Such knowledge could be described using an event graph. (c) A latent variable z is employed to learn the commonsense knowledge from event graph. tasks, as it is hard to express the complex semantics of natural language in a formal logic system. To facilitate this, Bhagavatula et al. (2019) proposed a natural language based abductive reasoning task αNLI. As shown in Figure 1 (a), given two ob- served events O1 and O2, the αNLI task requires the prediction model to choose a more reasonable explanation from two candidate hypothesis events H1 and H2. Both observed events and hypothe- sis events are daily-life events, and are described in natural language. Together with the αNLI task, Bhagavatula et al. (2019) also explored conducting such reasoning using pretrained language models such as BERT (Devlin et al., 2019) and RoBERTa (Liu et al., 2019). However, …

Content truncated to 500 words