Research on Generative Dialogue System Based on Reinforcement Learning

An open dialogue system model with diverse responses is constructed to try to solve the mono-tonous questions answered by the dialogue system during the response process. This paper proposes a generative dialogue method that combines bidirectional short-term memory neural network and reinforcement learning model. First, the corpus is preprocessed with various types of filters, so that the discourse corpus can be explored in a variety of ways; Secondly, in order to in-*