Tourism Chatbot Framework: Enhancing Visitor Experience Through GraphRAG and AI Chatbot

This study introduces a tourism chatbot frame-work based on the GraphRAG approach, designed to enhance visitor engagement through personalized recommendations for attractions, local products, and travel routes in Rayong, Thailand. By integrating Retrieval-Augmented Generation (RAG) with a tourism knowledge graph, the chatbot leverages both LLMs and structured data to deliver accurate, context-aware responses. The knowledge graph, built using Navanurak platform data, preserves Thailand's cultural heritage and biodiversity information, enriching the chatbots insights. The framework employs a hybrid search strategy, combining structured data retrieval and vector similarity search, enabling real-time, human-like interactions. Evaluation results show high precision, recall, and F1 scores for both location-based and general queries, demonstrating the chatbot's potential to promote lesser-known destinations and support local economies. This study highlights the effectiveness of integrating knowledge graphs with RAG to improve tourism experiences through enhanced information accessibility and personalization.