Association rule mining for mobile map personalisation

This paper presents an algorithmic solution to map personalisation for mobile users. Fundamental data filtering approaches are combined into a working system where content-based filtering is used for regular users who have their interests/preferences profiled and collaborative filtering is used for new/occasional users without user profiles. User map interactions are implicitly collected for user profile acquisition. Furthermore, association rule mining has been applied through the user map interactions to discover the association rules for geo-spatial features/services commonly accessed together. Such association rules are stored in a tree-like data structure for efficient storing and searching. Other commonly accessed features/services can be further recommended to the personalised map by collaborative filtering. Real world datasets have been used for our system and the initial system evaluation has shown promising.