The rapid growth of unstructured digital documents has created challenges in extracting relevant and accurate information efficiently. Traditional keyword-based search systems fail to understand semantic context and often return irrelevant results. Large Language Models have demonstrated strong capabilities in natural language understanding and generation; however, they may produce hallucinated or unverified responses when operating independently. This paper proposes an Intelligent Document Question Answering System that integrates Large Language Models with Retrieval-Augmented Generation to provide accurate, context- aware, and reliable answers grounded in source documents. The system processes documents by generating embeddings, storing them in a vector database, retrieving relevant content through similarity search, and generating responses using a language model. Experimental evaluation shows improved answer relevance, reduced hallucination, and enhanced retrieval accuracy compared to conventional search systems. The proposed system can be applied in enterprise knowledge management, legal analysis, healthcare documentation, and academic research
ABHINANTH et al. (Sun,) studied this question.