Jointly Identifying Rhetoric and Implicit Emotions via Multi-Task Learning

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 1429–1434 August 1–6, 2021. ©2021 Association for Computational Linguistics 1429 Jointly Identifying Rhetoric and Implicit Emotions via Multi-Task Learning Xin Chen1, Zhen Hai2, Deyu Li1,3, Suge Wang1,3∗ and Dian Wang1 1. School of Computer and Information Technology, Shanxi University, Taiyuan, China 2. R&D Center Machine Intelligence Technology Alibaba DAMO Academy, Singapore 3. Key Laboratory of Computational Intelligence and Chinese Information Processing of Ministry of Education, Shanxi University, Taiyuan, China cx simple@163.com, haiz0001@e.ntu.edu.sg {lidy,wsg}@sxu.edu.cn, sxu wdd@163.com Abstract Rhetorical implicit emotion identification is one of important and challenging tasks in nat- ural language processing. We observe that each rhetoric may express certain evidence of semantic and syntactic patterns. Then, we design a gate mechanism based classification module to capture respective rhetorical repre- sentation and identify each rhetoric. More- over, sentences carved with rhetoric tends to express emotions in subtle ways. We thus pro- pose a new multi-task learning framework that can encode the categorical correlation between tasks to improve the performance of rhetoric and emotion identification problem. Experi- mental results validate the benefit of the pro- posed model over state-of-the-art baselines for rhetoric and emotion identification tasks. In addition, a new Chinese rhetorical implicit e- motion dataset was constructed and will be re- leased in this work. 1 Introduction Rhetoric is formed by decorating the semantic- s and adjusting the structure of sentences in tex- t (Kennedy, 2009; Wang, 2013). The sentences that express implicit emotions via rhetorics are com- monly used in literary works and spiritual user re- views. They tend to express semantics at a high cognitive level and convey emotions in an obscure way. The example sentence below actually express- es the emotion of “love” and meanwhile consists of two rhetoric forms of parallelism and simile. 宽容是一块块木板，可以架起人与人之间的 桥梁；宽容是一根根丝带，可以系住人与人 之前的友谊；宽容是一团团火焰，可以融化 人与人之间的隔阂。 (Tolerance is a piece of board, which can build a bridge between people; Tolerance is a ribbon, which can tie the friendship ∗Corresponding author: Suge Wang. between people; Tolerance is a flame, which can melt the barriers between people.) Clearly, one can understand and predict the e- motion (“love”) and rhetoric (“simile” and “par- allelism”) of the sentence by capturing the se- mantic meanings of content words (“tolerance”, “board”, “bridge”, “ribbon”, “friendship”) as well as the syntactic structure and pattern of parallelism clauses in the sentence (“Tolerance is …, which can…”). Semantics and syntax are pivotal corner- stones for rhetoric, and are attached to different rhetoric. Moreover, positive emotions are more likely to be expressed by rhetorical categories such as simile, parallelism and so on, while negative emotions are more commonly conveyed by the rhetorical categories such as rhetorical question- s, irony and so on. Sentences carved with rhetoric can convey emo- tions in an implicit way, without explicitly using emotional words, which pose new challenges for textual understanding and rhetorical emotion anal- ysis in natural language processing. Various efforts have been made to cope with e- motion identification (Delbrouck et al., 2020; Chen et al., 2018; Klinger et …

Content truncated to 500 words