An Effective Program Clustering Algorithm for TV Recommendation System

As the increasing popularity of smart TV, providing interactive and personalized TV video recommendation to TV viewers as on internet has been an urgent requirements for TV operators. To recommend a program to viewer satisfying individual interest, the classification of TV programs will be necessary, which should reflect the users' interests. This paper presents an approach for clustering TV programs with using users' VoD(Video on Demand) data. After analyzing the VoD data source, the VoD duration ratio is proposed as the measure of user's interest to a TV program. Through analyzing clustering result, it is observed of the implicit relation of user interest and we explained the relation.