Abstract. In ubiquitous computing, behavior routine learning is the process of mining the context-aware data to find interesting rules on the user's behavior, while preference learning tries to utilize the user's behavior information to infer user interests, intention and desires. An intelligent environment should be adaptive, i.e. it is should be able to learn the routine and preference of user, then provide user with the suitable service. Developing intelligent ubiquitous environment requires not only good learning algorithms but also appropriate reusable models of user preference and behavior routine, which are not fully covered by current projects. In this paper, we propose a formal and comprehensive ontology-based model of user preference and behavior routine. The implementation of the ontology using OWL[14] enhances the expressiveness, support inference, knowledge reuse and knowledge sharing, which we can not achieve by normal models. The main benefit of this model is the ability to reason over context data to predict what the user wants the system to do. Based on our model, we also present a rule learning mechanism to learn the preference and behavior rules from context data.

No abstract

No abstract

No abstract