Summary: A Multi-Agent TV Recommender
Kaushal Kurapati, Srinivas Gutta, David Schaffer, Jacquelyn Martino, John Zimmerman
Adaptive Systems Department
Philips Research Briarcliff
345 Scarborough Rd.
Briarcliff Manor, NY 10510, USA
Personal Television is here via the advent of a new class of devices called personal video recorders (PVRs).
These recorders change the user task from (a) selecting a specific channel to watch from the 100+ available
channels to (b) finding something "good" to record from the 10,000+ shows broadcast each week.
Recommender systems, such as the one described in this paper, will help track users' preferences and aid
users in choosing shows to record. In this paper we advance a multi-agent TV recommender system that
encapsulates three user information streams--implicit view history, explicit preferences, and feedback
information on specific shows--into adaptive agents and generates program recommendations for a TV
viewer. We have tested the system in various agent combinations with real users drawn from a wide variety
of living conditions. The combination of implicit and explicit agents seems to work best in our framework.
Keywords: Multi-agent TV recommender system, user profiling, machine learning, user testing.