Leveraging hierarchical semantic‐emotional memory in emotional conversation generation

Abstract Handling emotions in human‐computer dialogues has emerged as a challenging task which requires artificial intelligence systems to generate emotional responses by jointly perceiving the emotion involved input posts and incorporating it into generation of semantically coherent emotionally reasonable responses. However, most previous works solely from posts, do not take full advantage training corpus suffer generating generic In this study, we introduce h ierarchical s emantic‐ e motional m emory module for c onversation (called HSEMEC), can learn abstract semantic conversation patterns information large corpus. The learnt knowledge helps enrich post representation assist generation. Comprehensive experiments on real‐world show that HSEMEC outperform strong baselines both automatic manual evaluation. For reproducibility, release code data publicly at: https://github.com/siat‐nlp/HSEMEC‐code‐data .