Research And Implementation Of Recommendation In Mobile TV Broadcast

With the advancements of Internet technology and rapid rising of mobile Internet, large amount of web resources are going to be presented to users through smartphone. When users enjoy the convenience brought by massive resources, they will also face the difficulty of too much selection. According to different users’characteristics, the personalized recommendation system can help improve user experience, optimize resource allocation and reduce uses’selection difficulty. This system has become a hot issue of current application research.As we all know, mobile TV broadcast is an extremely feature of mobile Internet. However, due to a lack of information and timeliness of TV shows, the existing recommenders performed poorly in this field. To tackle this issue, this dissertation carried out some related research based on current algorithms. The main work and achievement of this dissertation is as follows:1. Propose a real-time recommendation algorithm based on semantics and user preference in Live TV. Firstly, it constructs a user preference model based on the user’s watching history, and calculates the similarity between the to-be-recommended programs with the user’s watching history or the user’s watching programs respectively with the semantic similarity calculation method based on word vector. Combined with user preference, the similarity is used to work out the user’s virtual interest about the to-be-recommended programs. Finally, the program that gets the highest virtual interest becomes the user’s real-time recommendation result. Experimental results show this method to be 11% more effective than the current recommendation system when it’s used on the real-time prediction recommendation system. And the quality of the real-time program recommendation has been improved significantly by 15.1%.2. Propose a generation algorithm of virtual channels based on semantics and greedy selection. First of all, this method preprocessed the playbill and user’s viewing history to obtain intraday playbill of all channels and the user’s interest about watched TV shows. Afterwards, combined with the interest about corresponding program, it calculated the user’s virtual interest about programs in library. As the final step, it chose a program group with higher virtual interest, and generates a group of virtual channels which are continuous in time with the greedy algorithm. Experimental results show an improvement in the virtual channel’s profits by 30.6% and 21.6% compared with common channel and hot channel respectively.3. Designing and implementing a personalized mobile TV broadcast recommender system named "FengYun TV" with two recommendation methods mentioned above. This system recommends programs for users and it has two recommendation forms: real-time recommendation when users are watching living programs and personalized virtual channels. The system has been applied for Software copyright, and the corresponding mobile client is also uploaded to the Android application store for users.