Research on Text Generation of Medical Intelligent Question and Answer Based on Bi-LSTM and Neural Network Technology

Text generation is a popular research direction in the field of natural language processing as well as artificial intelligence, especially in the medical field. Text generation technology plays an extremely important role in medical intelligent question and answer systems. In previous Q&A systems, the display of generated dialogues is usually limited to Q&A and is difficult to adapt to vertical domains with strong characteristics. In this paper, to solve the task of specialized Q&A in medical field and provide users with comprehensive answers. We propose a TensorFlow architecture-based approach to medical text generation that uses sequential models in Keras to transform the task into a classification problem by treating the text generation problem as a prediction problem. By training and experimenting with the model on a large-scale Chinese medical question-and-answer dataset, the results show that our model has a good fit with applications in this specific domain. The model can aggregate medical domain knowledge, extract useful treatment information and generate medical knowledge text. At the same time, through research, it is found that splitting questions into phrases and then inputting them into the model can effectively solve the problem of repeated sentences in the generated answers. Since the model we proposed can answer patients' questions well and accurately in most cases, this research provides an improvement direction for exploring medical intelligent question answering and text generation.