A Semantic User Distance Metric Using GPS Trajectory Data

User similarity measure plays an important role in various location-based services including location prediction and recommendation. However, existing similarity computation methods fail to meet the distance metric axioms. In addition, existing works also suffer from some deficiency when identifying indoor stay regions and representing semantic information. To address these issues, this paper proposes a new method to evaluate user similarity by analyzing the global positioning system (GPS) trajectory data. Specifically, a more accurate algorithm for indoor stay region identification is proposed by taking velocity into account. Word embedding technique is used to compute the semantic distance between two stay regions. After that, stay regions are clustered and the user's GPS trajectories are represented as a multiset of semantic label sequences. Then a distance metric of these multisets satisfying the distance metric axioms is proposed on the basis of the balanced transportation problem. Finally, the effectiveness of the proposed method is evaluated by experiments using both synthetic and real-life datasets.

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.