Date of Publication :5th November 2024

Abstract:This paper introduces an advanced system to enhance PDF-based chatbots by integrating LangGraph, a powerful tool for constructing resilient language agents through a graph-based framework. The system improves information retrieval and response generation by utilizing an iterative feedback loop in conjunction with a Retrieval-Augmented Generation (RAG) model. By incorporating a relevancy threshold and ranking retrieved content, the system continuously adapts, learning from user feedback to refine its responses. The proposed approach ensures better retrieval accuracy and user experience across industries, where precise information extraction from complex documents is essential. The architecture is designed to overcome challenges associated with traditional IR systems, paving the way for innovative applications in various sectors.

Reference :