Key Issues and Theoretical Framework on Moving Objects Data Mining

摘要

Considering technical difficulties and bottlenecks in moving objects data mining, such as massive movement data, high dimensional data, topological complexity, and knowledge semantic representation etc., this paper focuses on the study of theory and methods of moving objects data mining. First, it presents two key scientific issues of the research topic, i.e. integration and modeling of heterogeneous data, and information aggregation and interpretation. Second, a theoretical framework of moving object data mining is proposed based on different perspectives of "spacetime data→space-time concept→space-time pattern". Two aspects of the framework are then discussed in details, including (1) moving objects data modeling and semantic expression;(2) mining methods and algorithms of association rules based on concept lattice. Finally, future works are discussed.