A dataset of knowledge injection for Gansu tourism large models

<p indent="0mm">With the digital transformation of the tourism industry, intelligent services have become an important part of modern travel. However, existing recommendation and Q&amp;A systems remain limited in terms of knowledge coverage and reasoning capabilities, making it difficult to meet diverse user demands. To this end, this study constructs a high-quality knowledge-injection dataset for tourism-oriented large models, focusing on the Gansu region and covering the three core dimensions: attractions, hotels, and cuisine. The dataset was constructed through multi-platform data collection, manual curation and strict quality control procedures, including data source selection, consistency checking and manual review, to ensure the accuracy, comprehensiveness and relevance of the data. After processing, the final dataset comprises 810,793 triples, 30,045 entities, and 14 relationship types, organized in the form of a knowledge graph. To facilitate intuitive display and exploration of multi-dimensional data associations, Neo4j is used as the storage and visualization tool. The release of this dataset provides strong data support for smart tourism and promote knowledge sharing and the development of intelligent systems in tourism domain.