Semantic Characterization of Context of Use and Contents for User-Centric Media Retrieval

When users access online media, they need and desire to get an experience tailored to their specific, personal context and situation. This is becoming more and more relevant with the ever-increasing amount of available contents users may choose from. In order to provide user-centric functionalities (such as relevant searches, content adaptation, customization and recommendation), both the annotation of contents with semantically rich metadata and an accurate model of the individual users and their respective contexts of use are needed. In this context, we propose a solution to automatically characterize both the context of use and the contents. It provides dynamic, adaptive user models, with explicit and implicit information; as well as content descriptors that may be later used to match the most suitable contents for each user. Users always keep a pivotal role throughout the whole process: providing new contents, contributing to moderated folksonomies, overseeing their own user model, etc.