The NoTube BeanCounter: Aggregating User Data for Television Programme Recommendation

In this paper we present our current experience of aggregat- ing user data from various Social Web applications and outline several key challenges in this area. The work is based on a concrete use case: reusing activity streams to determine a viewer's interests and generating television programme recommendations from these interests. Three sys- tem components are used to realise this goal: (1) an intelligent remote control: iZapper for capturing viewer activities in a cross-context televi- sion environment; (2) a backend: BeanCounter for aggregation of viewer activities from the iZapper and from dierent social web applications; and (3) a recommendation engine: iTube for recommending relevant tele- vision programmes . The focus of the paper is the BeanCounter as the rst