Azazel0203/Medical_ChatBot

# Technical Summary Extracts medical knowledge from PDF documents using RecursiveCharacterTextSplitter for chunking, then generates contextual embeddings via Hugging Face's sentence-transformers model and stores them in Pinecone for semantic similarity search. The Flask backend handles user queries by encoding them identically and retrieving the most relevant medical text chunks from Pinecone's vector index, enabling phrased-query matching beyond exact keyword matching. The system prioritizes scalability and medical domain accuracy, with provisions for swapping in domain-specific transformer models and integrating advanced conversation frameworks like Rasa for multi-turn dialogue handling.

No commits in the last 6 months.