RAG Chatbot for College Documents (Ask Your Syllabus)

Problem Statement & Abstract

Large language models are impressive, but they make things up when asked about documents they have never seen. Retrieval-Augmented Generation (RAG) fixes this by first finding the relevant passages and then asking the model to answer only from them. This project applies RAG to a college’s own documents.

Administrators upload PDFs such as syllabi, academic rules, exam notices and hostel guidelines. The system splits them into chunks, converts each chunk into an embedding vector and stores it in a vector database. When a student asks "What is the minimum attendance to sit the exam?", the most relevant chunks are retrieved and given to the language model, which writes a short answer with citations showing the document and page.

If the documents do not contain the answer, the bot says so instead of guessing. Conversations keep context for follow-up questions, answers stream word by word, and an evaluation script measures retrieval accuracy and answer faithfulness on a test set of questions. The model can be an open-source LLM running locally (via Ollama) or a hosted API.