Seq2seq Automatic Question Answering System of Medical Guide Station Based on Background Information

摘要

Automatic question answering technology brings great convenience to doctor-patient communication. Generally, a sequence-to-sequence (Seq2Seq) framework is used to build a question and answer model, but the model cannot make full use of the text information in the relevant context, and some answers generated are relatively simple. To this end, this paper combines medical background information with the Seq2Seq question and answer model, selects the independent recurrent neural network (IndRNN) as the codec of the question and answer model, and establishes an automatic question and answer system for the medical guidance station. Experiments have proved that the response generated by the Q&A model of the medical guidance platform that introduces medical background information is more rich and flexible, with higher accuracy, and close to the real medical guidance response.