Design of Intelligent Question Answering System Based on Large Language Model

With the extensive research of large language model, it has shown excellent natural language processing ability in the field of intelligent Q&A. However, there are still some problems in terms of accuracy and interpretability, as well as questions and answers in specific areas of expertise. Based on the relevant theories of LLM, this paper deeply studies the location coding, multi-head self-attention mechanism and layer standardization, laying a foundation for the application of ChatGPT. Based on ChatGPT 4.0 model, an intelligent Q&A system composed of basic data layer, structured data layer, intelligent question answering layer and front-end interaction layer is constructed and designed. In order to prove the correctness of ChatGPT 4.0 model selection, comparison experiments with other commonly used “ChatGPT 3.5, Claude 1.0 and ChatGLM3” models prove the advantages of ChatGPT 4.0 model in the field of intelligent question answering. Finally, the black-box test method is used to test the function of the system, which provides a basis for the system to enter the market and put into use.