Emotion-Cause Pair Extraction: A New Task to Emotion Analysis in Texts

Abstract

Emotion cause extraction (ECE), the task aimed at extracting the potential causes behind certain emotions in text, has gained much attention in recent years due to its wide applications. However, it suffers from two shortcomings: 1) the emotion must be annotated before cause extraction in ECE, which greatly limits its applications in real-world scenarios; 2) the way to first annotate emotion and then extract the cause ignores the fact that they are mutually indicative. In this work, we propose a new task: emotion-cause pair extraction (ECPE), which aims to extract the potential pairs of emotions and corresponding causes in a document. We propose a 2-step approach to address this new ECPE task, which first performs individual emotion extraction and cause extraction via multi-task learning, and then conduct emotion-cause pairing and filtering. The experimental results on a benchmark emotion cause corpus prove the feasibility of the ECPE task as well as the effectiveness of our approach.

A Mutually Auxiliary Multitask Model With Self-Distillation for Emotion-Cause Pair Extraction

“…Similar to the original ECPE methods [18], our work is still a two-stage based method: emotion and cause clauses are extracted in Stage…”

“…Inspired by multitask learning, Chen et al [38] studied jointly learning for emotion classification and emotion cause extraction, and proved the correlation between the two sub-tasks. Recently, Xia et al [18] further redefined this problem and proposed the ECPE task, aiming to extract the emotion and its causes in pairs. Tang et al [39] designed a joint model of emotion detection and emotion-cause pair extraction.…”

“…However, the ECE task needs to annotate emotions before extracting causes, which is very labor-consuming. To overcome this limitation, Xia et al [18] put forward a new task named emotion-cause pair extraction (ECPE), and proposed twostage methods which extract emotion clauses coupled with their cause clauses. As shown in Fig.…”

Suicidal Ideation Cause Extraction From Social Texts

“…To consider the relationships among sentences, they also proposed an emotion cause extraction framework (RTHN) (RNN-Transformer converter hierarchical network) [74]. To avoid manually annotating emotional clauses, they also proposed a joint extraction model of both emotional cause clauses and emotional clauses based on hierarchical BiLSTM and interactive multitask learning [75] and a joint extraction model based on a 2D square matrix.…”

Generative Adversarial Networks: A Literature Review

“…Natural language processing is the study of normal communication between humans and computers using natural language [78]. At present, NLP is mostly based on statistical machine learning and applied in emotional processing, machine translation, text extraction and other directions [79][80][81]. However, NLP did not make great progress in the early days of GAN.…”