Position Bias Mitigation: A Knowledge-Aware Graph Model for Emotion Cause Extraction

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pages 3364–3375 August 1–6, 2021. ©2021 Association for Computational Linguistics 3364 Position Bias Mitigation: A Knowledge-Aware Graph Model for Emotion Cause Extraction Hanqi Yan, Lin Gui, Gabriele Pergola, Yulan He Department of Computer Science, University of Warwick {hanqi.yan, lin.gui, gabriele.pergola, yulan.he}@warwick.ac.uk Abstract The Emotion Cause Extraction (ECE) task aims to identify clauses which contain emotion-evoking information for a particular emotion expressed in text. We observe that a widely-used ECE dataset exhibits a bias that the majority of annotated cause clauses are ei- ther directly before their associated emotion clauses or are the emotion clauses themselves. Existing models for ECE tend to explore such relative position information and suffer from the dataset bias. To investigate the degree of reliance of existing ECE models on clause rel- ative positions, we propose a novel strategy to generate adversarial examples in which the rel- ative position information is no longer the in- dicative feature of cause clauses. We test the performance of existing models on such adver- sarial examples and observe a significant per- formance drop. To address the dataset bias, we propose a novel graph-based method to ex- plicitly model the emotion triggering paths by leveraging the commonsense knowledge to en- hance the semantic dependencies between a candidate clause and an emotion clause. Ex- perimental results show that our proposed ap- proach performs on par with the existing state- of-the-art methods on the original ECE dataset, and is more robust against adversarial attacks compared to existing models.1 1 Introduction Instead of detecting sentiment polarity from text, recent years have seen a surge of research activities that identify the cause of emotions expressed in text (Gui et al., 2017; Cheng et al., 2017a; Rashkin et al., 2018; Xia and Ding, 2019; Kim and Klinger, 2018; Oberl¨ander and Klinger, 2020). In a typi- cal dataset for Emotion Cause Extract (ECE) (Gui 1Our code can be accessed at https://github.com /hanqi-qi/Position-Bias-Mitigation-in-Em otion-Cause-Analysis et al., 2017), a document consists of multiple clauses, one of which is the emotion clause an- notated with a pre-defined emotion class label. In addition, one or more clauses are annotated as the cause clause(s) which expresses triggering factors leading to the emotion expressed in the emotion clause. An emotion extraction model trained on the dataset is expected to classify a given clause as a cause clause or not, given the emotion clause. 1.71 7.71 54.45 23.58 7.47 2.22 0.51 0 10 20 30 40 50 60 Prev3 Prev2 Prev1 emotion Next1 Next2 Next3 Percentage(%) Cause8Position Figure 1: The distribution of positions of cause clauses relative to their corresponding emotion clauses in the ECE dataset (Gui et al., 2016). Nearly 87% of cause clauses are located near the emotion clause (About 55% are immediately preceding the emotion clause, 24% are the emotion clauses themselves and over 7% are imme- diately after the emotion clause). However, due to the difficulty in data collec- …

Content truncated to 500 words