Research On Personalized Service System For Social TV Users In Moblile Terminal

The development of social network has greatly changed people’s way of media consumption. More and more people turn to microblog for information of interest. Conversely, due to the fixed broadcast time and lack of interaction, etc, traditional TV is suffering from loss of subscribers and is thus actively looking forward to cooperate with social Network With such a tide of combination between traditional TV and social network, social TV—a new concept comes into the public eye.Compared with traditional TV, social TV is much more rich in its content. On one hand, the Internet offers much more kinds of television programs than before.On the other hand, social network also provides a new way to enjoy TV broadcasting. Users of microblogging service can instantly submit messages to express their opinions and feelings while watching TV. All of those tremendous explosion of information has made it hard for customers to find what they really need. Based on the above background, we focuse on the problem of personalized service in the field of social TV.Firstly, we study the technology of personalized TV program recommendation based on user-behavior data our system collected. A fine program model on program characteristics and an interest-granularity user model on users’watching-behaviors are proposed. In user modeling, we deeply discuss the quantization method of users’implicate feedback. Based on the above technologies, a hybrid program recommendation altorithm is proposed, combining content-based filtering algorithm and collaborative filtering algorithm.Secondly, we study the method of recommending useful TV-ralated microblog messages to the users of social TV. We solve the problem with a pipeline composed of two steps:message filtering and message classification. Considering the problem of title ambiguity and title lack, we caste the problem of TV-related message filtering into a problem of0-1 classification. As for message recommendation, influence factors of the message, impact of its author as well as the correlation between users and microblog messages, are all taken into account to find the queue of miroblog messages that match users’interest.At last, combination of two screens:mobile phone screen and TV screen is the mainstream realization way for social TV. Based on such idea, we designe and develope a social TV personalized service system. The server side of the system is mainly used to realize the model and algorithm mentioned above. The client side is a smartphone software, which is named as "TV assistant" and is responsible for user interation and behavior-data collection. We describe the system’s architecture, analysize the module design and presents the implementing solution of the system.