RAG-Based Patient Review Question-Answering Chatbot
Retrieval-Augmented Generation (RAG) system to answer questions from patient reviews. This repository contains a Jupyter notebook that demonstrates how to build a RAG-based chatbot for answering questions based on patient reviews.
LLM & Generative AI
- Google Gemini 2.5 Flash (Chat Model)
- LangChain (Prompt templates, Runnables, Chains)
- Google Generative AI Embeddings (
gemini-embedding-001)
Vector Database
- ChromaDB (persistent vector store)
- Similarity search retriever (k=10)
Data Handling & Preprocessing
- CSVLoader (for ingesting reviews.csv)
- Batched embedding generation with rate-limit safety
- Document structuring for retrieval modules
Prompting & RAG Pipeline
- ChatPromptTemplate
- System + Human message templates
- Context injection for hallucination-safe answers
- StrOutputParser for clean output formatting
Frontend / UI
- Gradio ChatInterface
- Real-time question answering
- Chat history & session-based interaction
UI Screenshot