Empathetic dialogue generation task aims at generating empathetic responses, based on perceived emotions instead of definite annotated emotions.

Current approaches mainly use an emotional label to generate an empathetic response for empathetic dialogue generation.

However, this method ignores the intrinsic emotional correlation in dialogue, making the generated response unsuitable.

This paper proposes a framework for enhancing empathetic dialogue generation with emotional correlation.

A multi-resolution emotion graph is designed to capture context-based emotion interactions from different resolutions.

Additionally, an emotion correlation-enhanced decoder generates a response based on correlation-aware aggregation and soft/hard strategy.

Powerpoint for this talk

Reference Paper

• Towards Empathetic Open-domain Conversation Models: A New Benchmark and Dataset
• Attention is all you need
• SKEP: Sentiment Knowledge Enhanced Pre-training for Sentiment Analysis