Chatbot System for Economic Data Analysis Using RAG and LangChain

This technical report presents a Proof of Concept (PoC) chatbot system that utilizes Retrieval-Augmented Generation (RAG) and LangChain technologies to facilitate interactive economic data analysis. The objective of the system is to democratize access to complex economic datasets, making it easier for a diverse user base, including policymakers, researchers, and the general public, to engage with authoritative economic information. The chatbot processes various economic datasets sourced from institutions such as the Bank for International Settlements (BIS) and the International Monetary Fund (IMF), and supports multiple data formats. The report details the methodology employed in developing the chatbot, including model selection, fine-tuning, and data source integration. It also outlines the embedding model used for semantic understanding of user queries and the retrieval mechanism that ensures accurate data matching. The integration of RAG enhances the reliability of insights by grounding responses in authoritative datasets, thereby improving the quality of economic data analysis.