Context-aware media recommendations for smart devices

Abstract

The emergence of pervasive computing, the rapid advancements in broadband and mobile networks and the incredible appeals of smart devices are driving unprecedented universal access and delivery of online-based media resources. As more and more media services continue to flood the Web, mobile users will continue to waste invaluable time, seeking content of their interest. To deliver relevant media items offering richer experiences to mobile users, media services must be equipped with contextual knowledge of the consumption environment as well as contextual preferences of the users. This article investigates context-aware recommendation techniques for implicit delivery of contextually relevant online media items. The proposed recommendation services work with a contextual user profile and a context recognition framework, using case base reasoning as a methodology to determine user's current contextual preferences, relying on a context recognition service, which identifies user's dynamic contextual situation from device's built-in sensors. To evaluate the proposed solution, we developed a case-study context-aware application that provides personalized recommendations adapted to user's current context, namely the activity he/she performs and consumption environment constraints. Experimental evaluations, via the case study application, real-world user data, and online-based movie metadata, demonstrate that context-aware recommendation techniques can provide better efficacy than the traditional approaches. Additionally, evaluations of the underlying context recognition process show that its power consumption is within an acceptable range. The recommendations provided by the case study application were assessed as effective via a user study, which demonstrates that users are pleased with the contextual media recommendations.