Research On Dynamic TV Program Recommendation Method For Family Users

In recent years,with the rapid development of digital television and communication technology,television programs are more and more colorful,the emergence of a large number of television programs lead to the problem that it is difficult for television viewers find their own interest in television programs and in a certain extent affect the TV Program ratings.At present,recommending television programs for TV viewers interested in have become a common need for TV service providers and television viewers,and are also one of the important issues in the area of recommendation system.First,this paper analyzes the characteristics and challenges of TV program recommendation system,studies the current situation of user interest preference modeling and recommendation for family users,introduces the typical recommendation algorithm and evaluation index of TV program recommendation system,provides the theoretical and practical basis for research work in user interest preference modeling for TV viewers and personalized recommendation method for family users.Then,aiming at the issue of hard to obtain user explicit preferences in Personalized TV program recommendation,we propose an approach of building user interest preference model based on user viewing behavior.This approach not only uses the operation and duration in the process of viewing TV programs,but also combines with the basic properties of TV programs,then describes user interest preference from two dimensions of time and frequency,and establishes interest preference matrix.The experimental results show that the proposed approach can describe user interest preference accurately and perform well in precision and recall.Finally,with the problem of Personalized TV program recommendation generally face diverse family users,we propose a combination of offline recommendation and online real-time update recommended method.This method not only takes into account the preferences of the whole family users to meet the needs of individual family members,but also can effectively dig out the current user’s interest preferences according to the real-time behavior of the current family members.The experimental results show that the method can effectively identify the preferences of the current family members and can put the current family favorite TV show standing in the front.