Personalized recommendation of live programs in cable television

Aiming at reducing the overload of cable television information and the difficulty to pinpoint users, this research employs the directed preview method of CATV's live programs which is grounded in labels. Long-term and short-term models of users' viewing behaviors and models of period preference are constructed accurately and can multi-dimensionally describe users' viewing behaviors. Based on labels, the research propounds the TOPN method of recommendation of live program, which could increase the accuracy rate of recommendation to 40%.