STCCD: Semantic trajectory clustering based on community detection in networks. (30th December 2020)

• Record Type: Journal Article
• Authors: Liu, Caihong; Guo, Chonghui
• Is Part Of: Expert systems with applications. Volume 162(2020)
• Publication Date: 2020-12-30
• Subjects: Trajectory clustering -- Trajectory similarity -- Complex network -- Community detection; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2020.113689
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: Proposing a framework for semantic trajectory clustering based on community detection. Introducing the ontology theory to semantic trajectory similarity computation. Constructing a trajectory network according to the semantic trajectory similarity. Conducting extensive studies and analyses on three real world trajectory datasets. Comparing the proposed method with some traditional and recently proposed methods. Abstract: Most of traditional trajectory clustering algorithms often cluster similar trajectories from a temporal or spatial perspective. One weak point is that the semantic relationship between the trajectories is ignored. In some cases, trajectories with spatio-temporal similarities may be semantically related, and the negligence of semantic information may result in unreasonable trajectory clustering results. In addition, the existing semantic trajectory clustering algorithms only consider the local semantic relationship between adjacent spatio-temporal trajectories, and the overall global semantic relationship between trajectories is still unknown. Considering the disadvantages of the current trajectory clustering methods, we proposed a novel algorithm for semantic trajectory clustering based on community detection (STCCD) in networks, which can better measure the semantic similarity of trajectories and capture global relationship among trajectories from the perspective of the network, and can get better trajectory clustering results compared to some traditional and recently proposed methods. Experimental results demonstrate that the proposed method can effectively mine the trajectory clustering information and related knowledge from the semantic trajectory data.