AI-BASED COLLEGE INFORMATION ASSISTANT USING RAG

Students often need to search across college websites, admission brochures, circulars, academic calendars, fee notices, and
departmental documents to find reliable information. This paper proposes an AI-based College Information Assistant using
Retrieval-Augmented Generation (RAG), a method that combines information retrieval with language generation. The proposed
system accepts a natural-language question, retrieves relevant passages from approved college documents, and uses those
passages to generate a concise response with source references. The workflow includes document ingestion, text cleaning,
chunking, embedding generation, semantic retrieval, grounded answer generation, and safe fallback handling when evidence is
missing or conflicting. The design emphasizes traceability, document freshness, privacy, and avoidance of unsupported claims. A
suggested evaluation plan measures retrieval relevance, answer correctness, citation support, abstention quality, response time,
and usability. The paper describes the problem, related work, architecture, methodology, implementation options, test scenarios,
limitations, and future scope. As no implemented prototype or measured dataset results were supplied, the paper does not claim
numerical performance; these must be added after actual testing