Construction of Multi-modal Chinese Tourism Knowledge Graph

EasyChair Preprint 4013

Abstract

This paper proposes a construction process of Multi-modal Chinese Tourism Knowledge Graph (MCTKG). In order to maintain the semantic consistency of heterogeneous data sources, the tourism ontology is constructed by semi-automatic method, and the data sources are aligned to the tourism ontology. The construction process of MCTKG includes following modules: ontology construction, entity alignment, tourism route automatic generation, sharing platform establishment. In ontology construction, semi-automatic fine-tuning operations are carried out, which optimize and simplify the concept hierarchy relationship abstracted from the obtained data resources, and some new concept hierarchy relationships are manually defined according to the actual application scenarios. In entity alignment, we adopt a method based on n-gram distance to align entities from different sources, and fuse and cross-validate their attributes. In addition, based on the concepts of attraction and tourism style in the knowledge graph, we propose a tourism route generation algorithm, which could automatically schedule the tourism routes by incorporating the characteristics of attraction and tourists’ demands. Lastly, a sharing platform is established, which provides open data access and query interface.

Keyphrases: Entity Alignment, Knowledge Graph, Tourism Ontology, route generation

@booklet{EasyChair:4013,
  author    = {Xie and Qinghua Wen and Hailong Jin and Zhenhao Dong and Lei Hou and Hongyin Zhu and Juanzi Li},
  title     = {Construction of Multi-modal Chinese Tourism Knowledge Graph},
  howpublished = {EasyChair Preprint 4013},
  year      = {EasyChair, 2020}}