Emotion Dynamics Modeling via BERT

TL;DR

This paper introduces BERT-based models for emotion dynamics in conversations, effectively capturing dependencies among interlocutors and improving emotion recognition accuracy over previous RNN-based methods.

Contribution

It develops novel BERT-based architectures, including F-BERT, H-BERT, and ST-BERT, to better model conversational emotion dynamics and influence among interlocutors.

Findings

Achieved around 5% and 10% improvements on benchmark datasets.

Demonstrated effectiveness of BERT-based models over RNNs.

Enhanced modeling of inter- and intra-interlocutor dependencies.

Abstract

Emotion dynamics modeling is a significant task in emotion recognition in conversation. It aims to predict conversational emotions when building empathetic dialogue systems. Existing studies mainly develop models based on Recurrent Neural Networks (RNNs). They cannot benefit from the power of the recently-developed pre-training strategies for better token representation learning in conversations. More seriously, it is hard to distinguish the dependency of interlocutors and the emotional influence among interlocutors by simply assembling the features on top of RNNs. In this paper, we develop a series of BERT-based models to specifically capture the inter-interlocutor and intra-interlocutor dependencies of the conversational emotion dynamics. Concretely, we first substitute BERT for RNNs to enrich the token representations. Then, a Flat-structured BERT (F-BERT) is applied to link up utterances in a conversation directly, and a Hierarchically-structured BERT (H-BERT) is employed to distinguish the interlocutors when linking up utterances. More importantly, a Spatial-Temporal-structured BERT, namely ST-BERT, is proposed to further determine the emotional influence among interlocutors. Finally, we conduct extensive experiments on two popular emotion recognition in conversation benchmark datasets and demonstrate that our proposed models can attain around 5\% and 10\% improvement over the state-of-the-art baselines, respectively.