ConText at WASSA 2024 Empathy and Personality Shared Task: History-Dependent Embedding Utterance Representations for Empathy and Emotion Prediction in Conversations

TL;DR

This paper presents a context-aware embedding approach for predicting empathy and emotion in conversations, leveraging historical dialogue data with pre-trained language models, achieving top rankings in WASSA 2024 shared tasks.

Contribution

It introduces a history-dependent embedding method that incorporates conversational context and speaker identification for improved empathy and emotion prediction.

Findings

Ranked 1st in CONV-turn track

Ranked 2nd in CONV-dialog track

Effective modeling of conversational context enhances prediction accuracy

Abstract

Empathy and emotion prediction are key components in the development of effective and empathetic agents, amongst several other applications. The WASSA shared task on empathy and emotion prediction in interactions presents an opportunity to benchmark approaches to these tasks. Appropriately selecting and representing the historical context is crucial in the modelling of empathy and emotion in conversations. In our submissions, we model empathy, emotion polarity and emotion intensity of each utterance in a conversation by feeding the utterance to be classified together with its conversational context, i.e., a certain number of previous conversational turns, as input to an encoder Pre-trained Language Model, to which we append a regression head for prediction. We also model perceived counterparty empathy of each interlocutor by feeding all utterances from the conversation and a token identifying the interlocutor for which we are predicting the empathy. Our system officially ranked $1^{st}$ at the CONV-turn track and $2^{nd}$ at the CONV-dialog track.