Research on Man-Machine Conversation System Based on GRU seq2seq Model

Abstract

Intelligent robot Q & A is a new focus in the field of natural language understanding and processing. In recent years, with the gradual maturity of deep learning technology, more and more research has begun to model deep neural networks for text semantics. Deep neural networks are computational models abstracted from human brain cognitive processes, closer to humans. In this paper, we use GRU deep neural network to study the generative dialogue, and implement a seq2seq model manmachine conversation system. Experiments show that the confidence of the system can reach the range of human acceptance.