PRCCF: A Persona-guided Retrieval and Causal-aware Cognitive Filtering Framework for Emotional Support Conversation

TL;DR

PRCCF is a novel framework for emotional support conversation that combines persona-guided retrieval and causality-aware filtering to improve empathetic response generation.

Contribution

It introduces a new framework integrating persona-guided retrieval and causality-aware filtering to enhance emotional understanding in conversation models.

Findings

PRCCF outperforms state-of-the-art baselines on ESConv dataset.

The framework improves both automatic metrics and human evaluation scores.

Causality-aware filtering enhances contextual cognitive understanding.

Abstract

Emotional Support Conversation (ESC) aims to alleviate individual emotional distress by generating empathetic responses. However, existing methods face challenges in effectively supporting deep contextual understanding. To address this issue, we propose PRCCF, a Persona-guided Retrieval and Causality-aware Cognitive Filtering framework. Specifically, the framework incorporates a persona-guided retrieval mechanism that jointly models semantic compatibility and persona alignment to enhance response generation. Furthermore, it employs a causality-aware cognitive filtering module to prioritize causally relevant external knowledge, thereby improving contextual cognitive understanding for emotional reasoning. Extensive experiments on the ESConv dataset demonstrate that PRCCF outperforms state-of-the-art baselines on both automatic metrics and human evaluations. Our code is publicly available at: https://github.com/YancyLyx/PRCCF.