An Ontology-based Approach for Mobile Personalized Recommendation

Mobile devices are in a widespread use today. Restricted by the features of mobile devices, such as mobility, low communication bandwidth, small capacity of memory, limited power and inconvenient interaction, mobile services are required to recommend items adapting to the user need and location. Much existing research on recommendations ignores semantic information of user dynamic historical preferences which is vital to the quality of recommendation. In this paper, an ontology-based approach for mobile personalized recommendation is proposed. This paper gives weight values to historical preferences using a decay factor and a time window. Then, historical preferences are mapped into the domain ontology and generate the User Dynamic Preference Ontology (UDPO), which is stored in mobile devices to reduce the load of bandwidth. After building UDPO, the user profile, the domain ontology and UDPO are combined to calculate the set of recommended items. Here, the user profile involves the user location and the user interests. The experimental results show that our approach can provide Points of Interest which the user likes better.