Trajectory data mining: integrating semantics

摘要

Purpose - Trajectory is the path a moving object takes in space. To understand the trajectory movement patters, data mining is used. However, pattern analysis needs semantics to be understood. Therefore, the purpose of this paper is to enrich trajectories with semantic annotations, such as the name of the location where the trajectory has stopped, so that the paper is able to attain quality decisions. Design/methodology/approach - An experiment was conducted to explain that the use of raw trajectories alone is not enough for the decision-making process and detailed pattern extraction. Findings - The findings of the paper indicates that some fundamental patterns and knowledge discovery is only obtainable by understanding the semantics underlying the position of each point. Research limitations/implications - The unavailability of data are a limitation of the paper, which would limit its generalizability. Additionally, the lack of availability of tools for automatically adding semantics to clusters posed as a limitation of the paper. Practical implications - The paper encourages governments as well as businesses to analyze movement data using data mining techniques, in light of the surrounding semantics. This will allow, for example, solving traffic congestions, since by understanding the movement patterns, the traffic authority could make decisions in order to avoid such congestions. Moreover, it could also help tourism authorities, at national levels, to know tourist movement patterns and support these patterns with the required logistical support. Additionally, for businesses, mobile operators could dynamically enhance their services, voice and data, by knowing the semantically enriched patterns of movement. Originality/value - The paper contributes to the already rare literature on trajectory mining, enhanced with semantics. Mainstream literature focuses on either trajectory mining or semantics, therefore the paper claims that the approach is novel and is needed as well. By integrating mining outcomes with semantic annotation, the paper contributes to the body of knowledge and introduces, with lab evidence, the new approach.