Event-centric Tourism Knowledge Graph - A Case Study of Hainan.

Jie Wu,Xinning Zhu,Chunhong Zhang,Zheng Hu

DOI: https://doi.org/10.1007/978-3-030-55130-8_1

2020-01-01

Abstract:Knowledge graphs have become increasingly important in tourism industry recently for their capability to power insights for applications like recommendations, question answering and so on. However, traditional tourism knowledge graph is a knowledge base which focuses on the static facts about entities, such as hotels, attractions, while ignoring events or activities of tourists' trips and temporal relations. In this paper, we first propose an Event-centric Tourism Knowledge Graph (ETKG) in order to model the temporal and spatial dynamics of tourists trips. ETKG is centered on activities during the trip and regards tourists' trajectories as carriers. We extract valuable information from over 18 thousand travel notes crawled from Internet, and define an ETKG schema to model tourism-related events and their key properties. An ETKG based on touristic data in Hainan is presented which incorporates 86977 events (50.61% of them have complete time, activity, location information) and 7132 journeys. To demonstrate the benefits of ETKG, we propose an Event-centric Tourism Knowledge Graph Convolutional Network (ETKGCN) for POI recommendations, which facilitates incorporating tourists behavior patterns obtained from ETKG, so as to capture the relations between users and POIs more efficiently. The offline experiment results show that our approach outperforms strong recommender baselines, so that it validates the effectiveness of ETKG.