LLM-based user requirement analysis and intelligent Q&A toward Chang’an Twelve Hours

The new technology of artificial intelligence can bring users a fresh experience and accelerate the rapid development of cultural and tourism integration. In this paper, we use a large language model (LLM) to mine user implicit demands and implement intelligent Q&A reasoning for the Chang’an Twelve Hours Scenic Area. First, multi-faceted data related to Chang’an Twelve Hours is collected to construct a vector database. Next, the LLM is utilized to mine the implicit requirements of user questions by combining attribute features such as gender and age. Then, the user demand vectors are retrieved and matched with the content vectors in the vector database. Finally, the similar content obtained from user demand and retrieval feedback is used to populate the prompt template and input into the LLM, realizing intelligent Q&A and reasoning for the Twelve Hours of Chang’an. On this basis, a prototype of an intelligent Q&A system tailored to the Twelve Hours of Chang’an is constructed. Experimental comparisons with the traditional Retrieval-Augmented Generation (RAG) technique demonstrate that using an LLM enables finer mining of user requirements and generates more accurate answers. The intelligent Q&A system for Chang’an Twelve Hours developed in this paper effectively addresses users’ customized inquiries, thereby enhancing the user experience and improving the operational efficiency of the scenic spot. Additionally, this system can serve as a reference model for implementation at other cultural tourism attractions.