GraphLoc: a graph based approach for automatic detection of significant locations from GPS trajectory data. Issue 1 (2nd January 2018)

• Record Type: Journal Article
• Authors: Shams, Bita; Haratizadeh, Saman
• Is Part Of: Journal of spatial science. Volume 63:Issue 1(2018)
• Page Start: 115
• Page End: 134
• Publication Date: 2018-01-02
• Subjects: Location discovery -- local properties -- sharp boundary problem -- Graph; Geographic information systems -- Periodicals; Cartography -- Periodicals; Surveying -- Periodicals; Geodesy -- Periodicals; Photogrammetry -- Periodicals; Cartography; Geodesy; Geographic information systems; Photogrammetry; Surveying; Periodicals; 526.05
• DOI: 10.1080/14498596.2017.1327374
• Languages: English
• ISSNs: 1449-8596

Abstract

Abstract: Automatic discovery of significant locations from row GPS data is the first phase of mining mobility pattern and developing location-aware services. Unfortunately, current location discovery algorithms are ineffective when locations have different local properties such as density. Moreover, these algorithms suffer from the sharp boundary problem that is assigning some close points to different locations while they intuitively belong to one location. This article presents a novel framework, GraphLoc that formulates location discovery as a network community detection problem to address these issues. Experimental results show that GraphLoc's locations lead to higher performance in mobility mining and location prediction.