Smart media recommender system based on semi supervised machine learning

Predicting user preferences and providing personalized services based on his past preferences present an important issue in the field of pervasive computing. However, studies considering users' preferences are relatively insufficient in this domain. The aim of this paper is to propose an approach to provide personalized services to users, using context history and machine learning techniques. In this approach, we integrate, to pervasive recommender systems, the ability of predicting user preferences on new context situations even in unforeseen contexts that have not been considered when building the knowledge base of the system. And this, in order to serve the user in a proactive and uninterrupted way in various contexts that may arise in the future.