Single turn Chinese emotional conversation generation based on information retrieval and question answering

Emotional conversation is attracting more attention in natural language understanding and machine intelligence. In this paper, we propose a single turn Chinese emotional conversation generation system. Given a post, our system generates appropriate responses from large-scale conversation datasets together with corresponding emotional labels. This system consists of four components, i.e., information retrieval (IR) system, language model, question answering (QA) system and emotion classifier. The experimental results on benchmark dataset show that our system is capable of retrieving appropriate responses with emotional labels.