Mixed Knowledge-enhance Empathetic Dialogue Generation

Abstract

Empathy plays a pivotal role in human communication, and thus, it is an essential capability that any human-centered dialogue system should possess. Early research on empathetic response generation often focused on directly capturing the emotional state of the context using fixed emotion labels. However, the logical aspects exhibited in human conversations heavily rely on experiential and knowledge-based resources within the brain. This implies that whether the aim is to acquire more nuanced emotional states or to generate responses enriched with comprehensive information, the incorporation of external knowledge as supplementary information in empathetic dialogue systems is imperative. In response to this challenge, we propose a novel approach for extracting external knowledge. This is achieved by designing two components: a fine-grained knowledge graph constructed using the context and an external knowledge base, and coarse-grained knowledge acquisition based on COMET. These two scales of knowledge are then integrated with the context using methods like context refinement. This not only make the model to gain a deeper understanding of the user's context but also enhances the expression of empathy in the dialogue system. We conducted extensive experiments on the EMPATHETICDIALOGUES dataset and demonstrated the superiority of our approach over the baseline model.