Recommender systems are a means of personalisation providing their users with personalised recommendations of items that would
possibly suit the users needs. They are used in a broad area of contexts where items are somehow linked to users. The creation
of recommendations of interactive live TV suffers from several inherent problems, e.g. the impossibility to foresee the contents
of the next items or the reactions of the user to the changing programme.
This paper proposes an algorithm for building personalised streams within interactive live TV. The development of the algorithm
comprises a basic model for users and media items. A first preliminary evaluation of the alogithm is executed and the results
discussed.
Keywords recommender system - interactive live TV - multistream