Seminar: IPTV Recommender Systems

Speaker: Prof. Paolo Cremonesi,

Computer Science Department, Politecnico di Milano, Italy.

Date/Time: Note: The date has now changed to 28th June 2007, 13:00-14:00.

Location: Room G74, Philip Lyle Building, Whiteknights Campus.

Map: http://www.info.rdg.ac.uk/maps/maps-display.asp

Abstract:

Nowadays the amount of content (books, movies, music, etc.) that is available overcomes our processing abilities. This problem is particularly relevant for IPTV users, because of the difficulties in browsing the large number of programs using a remote controller.

In the last few years, recommender systems have emerged as a powerful technique for dealing with this issue. Such systems recommend the user on the basis of a model of the user profile, i.e., her/his tastes, interests or attitudes. How much an item interests a user is called relevance. Thus, a recommender system estimates the relevance of a user on a collection of items. To date, most recommendation techniques have been designed to recommend books, DVDs and CDs to on-line users of e-commerce sites.

Designing effective recommender systems for IPTV infrastructures is more problematic than its equivalent on web-based e-commerce sites because of a number of technological problems. IPTV infrastructures exhibit an almost unlimited number of items (e.g., movies, channels, TV programs) continuously growing over time. The consequence is that each item gets less and less visibility and users are scarcely stimulated in purchasing them.

This talk will explore the extension of recommender system to the IPTV domains and will highlight some interesting research challenges that could advance both the usefulness and broad deployment of personalization techniques.