A Multilingual Entity Alignment-Driven Xinjiang Tourism Knowledge Graph Dataset

With the deepening of international tourism exchanges driven by the ‘Belt and Road’ initiative, this paper constructs MultiXJ-Tourism, a multilingual cultural and tourism knowledge graph dataset for cross-language intelligent service scenarios in Xinjiang, which covers the core elements of attractions, food, shopping, entertainment, and accommodations, and contains a total of 911409 triples, 100417 entities, and 1520 types of relationships. , 81,878 entities and 18 types of relationships. The dataset adopts structured graphical representation and supports multi-language alignment in Chinese, English, Russian and Uyghur, which significantly improves the cross-language understanding and information retrieval ability of the tourism model. The study proposes a three-stage semantic alignment strategy of ‘translation-alignment-verification’ to unify the multilingual entity modelling, and provides fine-grained attribute annotation to support accurate Q&A and recommendation services. This dataset is suitable for intelligent Q&A, multilingual model training and cross-language retrieval, and provides high-quality data support for cultural tourism promotion and regional research in the context of ‘One Belt, One Road’, and this research not only fills the gaps of the multilingual tourism knowledge map in ethnic minority areas, but also provides an important tool for the research of knowledge enhancement of multilingual tourism large models and multilingual knowledge representation. This research not only fills the gap of multilingual tourism knowledge mapping in minority areas, but also provides an important example for knowledge enhancement and multilingual knowledge representation in multilingual tourism.