RAG Chatbot for Policy Document Enquiry and Feedback

In today’s era of rapidly evolving digital ecosystems, accessing and understanding large, complex policy documents remains a major challenge for students, institutions, and stakeholders. Traditional manual search methods are inefficient, time-consuming, and prone to misinterpretation. To overcome this limitation, this paper presents a Retrieval-Augmented Generation (RAG) powered chatbot that integrates large language models (LLMs) with similarity-based retrieval mechanisms for precise and context-aware responses. By leveraging a vector store and embeddings for document chunking, the chatbot retrieves the most relevant sections of the policy and generates human-like answers using the Grok LLM. This hybrid approach ensures factual consistency, improves user accessibility, and provides real-time feedback collection. Experimental evaluation demonstrates that the RAG chatbot significantly improves query accuracy, reduces response time, and enhances user satisfaction. The system shows strong potential for scalable, cost-effective, and interactive policy assistance in academic and administrative domains.