Automatic Point of Interests Detection Using Spatio-Temporal Data Mining Techniques over Anonymous Trajectories

摘要

Location Based Services (LBS) are evolving very fast but despite this are still at an early phase of their evolution. The future LBS services will be more anticipatory of users needs and will exploit a broader range of information on users. In order to anticipate users' needs, LBS providers should be able to understand users' behaviours, preferences and interests automatically and without asking the user to specify them. Then using the user's current situation and previously extracted behaviours, interests and dislikes, the user's needs can be predicted at the relevant moment and provide the most appropriate sets of services. This paper shows the application of data mining techniques over anonymous sets of tracking data to recognise mobility behaviours and preferences of users such as Point of Interest (PoI). Tracking data are first anonymised then stored in a spatio-temporal database. Then, using data mining techniques, rules, models and patterns are recognised. Such knowledge, patterns and models are subsequently used for intelligent navigation services as input making suggestions.