TED: Turn Emphasis with Dialogue Feature Attention for Emotion Recognition in Conversation

TL;DR

This paper introduces TED, a novel attention mechanism that explicitly incorporates dialogue features like turn position and speaker info to improve emotion recognition in multi-turn conversations, achieving state-of-the-art results.

Contribution

The paper proposes a priority-based attention method called TED that explicitly models dialogue features to enhance emotion recognition in conversation.

Findings

TED outperforms existing methods on four benchmark datasets.

Achieves state-of-the-art performance on IEMOCAP with many turns.

Demonstrates high overall performance across all evaluated datasets.

Abstract

Emotion recognition in conversation (ERC) has been attracting attention by methods for modeling multi-turn contexts. The multi-turn input to a pretraining model implicitly assumes that the current turn and other turns are distinguished during the training process by inserting special tokens into the input sequence. This paper proposes a priority-based attention method to distinguish each turn explicitly by adding dialogue features into the attention mechanism, called Turn Emphasis with Dialogue (TED). It has a priority for each turn according to turn position and speaker information as dialogue features. It takes multi-head self-attention between turn-based vectors for multi-turn input and adjusts attention scores with the dialogue features. We evaluate TED on four typical benchmarks. The experimental results demonstrate that TED has high overall performance in all datasets and achieves state-of-the-art performance on IEMOCAP with numerous turns.