Summarize before Aggregate: A Global-to-local Heterogeneous Graph Inference Network for Conversational Emotion Recognition

Proceedings of the 28th International Conference on Computational Linguistics, pages 4153–4163 Barcelona, Spain (Online), December 8-13, 2020 4153 Summarize before Aggregate: A Global-to-local Heterogeneous Graph Inference Network for Conversational Emotion Recognition Dongming Sheng1,2 Dong Wang1,2 Ying Shen3 Hai-Tao Zheng1,2,∗ Haozhuang Liu1,2 1Department of Computer Science and Technology, Tsinghua University 2Shenzhen International Graduate School, Tsinghua University 3School of Intelligent Systems Engineering, Sun Yat-Sen University {cdm19, wangd18, lhz19}@mails.tsinghua.edu.cn sheny76@mail.sysu.edu.cn, zheng.haitao@sz.tsinghua.edu.cn Abstract Conversational Emotion Recognition (CER) is a crucial task in Natural Language Processing (NLP) with wide applications. Prior works in CER generally focus on modeling emotion influ- ences solely with utterance-level features, with little attention paid on phrase-level semantic con- nection between utterances. Phrases carry sentiments when they are referred to emotional events under certain topics, providing a global semantic connection between utterances throughout the entire conversation. In this work, we propose a two-stage Summarization and Aggregation Graph Inference Network (SumAggGIN), which seamlessly integrates inference for topic-related emo- tional phrases and local dependency reasoning over neighbouring utterances in a global-to-local fashion. Topic-related emotional phrases, which constitutes the global topic-related emotional connections, are recognized by our proposed heterogeneous Summarization Graph. Local de- pendencies, which captures short-term emotional effects between neighbouring utterances, are further injected via an Aggregation Graph to distinguish the subtle differences between utter- ances containing emotional phrases. The two steps of graph inference are tightly-coupled for a comprehensively understanding of emotional fluctuation. Experimental results on three CER benchmark datasets verify the effectiveness of our proposed model, which outperforms the state- of-the-art approaches. 1 Introduction Conversational Emotion Recognition (CER) has attracted increasing interests for its promising applica- tions in intelligent interactive systems with diverse functionalities, including medical-care systems and online recommendation systems (Zhang et al., 2014; Gkotsis et al., 2016; Chen et al., 2018; Shen et al., 2020). As shown in Figure 1, conversations in CER datasets are segmented into multiple utterances based on breaths or pauses of the speaker, and each utterance is associated with an emotion label. Existing works with deep learning approaches generally capture emotional features solely from an utterance-level perspective by modeling interactions between utterances via Recurrent Neural Network (RNN) structures (Hazarika et al., 2018b; Hazarika et al., 2018a; Majumder et al., 2019) or graph struc- tures (Ghosal et al., 2019). However, phrase-level semantic connection between utterances is still under- explored, creating a substantial barrier for comprehensively understanding of the source of emotional fluctuation. An example of a topic-related emotional phrase is shown in Figure 1, where the noun phrase “fifty dollars” is not an emotional expression at first, but under the topic of compensation, it has been associated with anger and frustration by referring to the unsatisfactory compensation. Therefore, it is crucial to recognize phrase-level emotional patterns globally across different utterances. Furthermore, topic-related emotional phrases generally scatter in different utterances throughout the conversation, and the emotions they convey depend on local context. To accurately distinguish emotions behind these “fifty dollars” phrases, the model is required to be comprehensively aware of both the topic about com- pensation and …

Content truncated to 500 words