Design and Implementation of a Medical Question Answering System Based on Retrieval-Augmented Generation

With the rapid growth of medical information demand, providing accurate and reliable medical question-answering services has become a pressing challenge. Traditional generative Q&A models are prone to hallucinations, frequently generating inaccurate medical advice, while simple retrieval systems lack natural fluency. To address this, we propose a Retrieval-Augmented Generation (RAG)-based medical QA system with a five-layer architecture. The data processing layer cleans and formats data from the MedChatZH dataset. The knowledge storage layer encodes documents into 768-dimensional vectors using the BGE-small-zh-v1.5 embedding model and builds an index. The retrieval layer selects relevant documents via cosine similarity. The answer generation layer employs the ChatGLM3-6B language model for context-aware responses. Finally, a user-friendly web interface is implemented using Gradio, offering high-quality medical Q&A services.