Implementation Of Set-Top-Box Program Recommendation System Based On Interest Computing Of RS-TS Model

With the proposal of the concept of smart home, smart TV set-top boxes began to enter millions of households. Currently more representative of the TV set-top boxes, including Apple’s iTV, XiaoMiTV box,LeTV csl and the PPBoX TV box of the PPTV company and so on. Various TV box core function is to provide a wealth of video sources, support for users based on their viewing preferences on-demand and customized programs. How to recommend the content that users really like from massive program determines the TV box product’s user experience, it’s related to the product reputation and market share.Interest computing is a subset of social computing. Simply speaking, the main work is computing the historical record and user behavior analysis, mining the user’s interest, and to satisfy the individual needs of each user, and then improving the user’s experience. Interest computing in the recommended system is highly targeted, can provide relatively accurate recommendations set.In this paper, to solve these problems, the calculation of interest based on social computing concepts presented RS-TS model that is the Real Satisfaction-Temptation Satisfaction model, combined with the amount of time decay and click adjustment algorithm to achieve a complete STB program recommendation system.This paper designed and implemented the recommendation system that based on the interest calculation and RS-TS model on STB, the system can be divided into two parts, statistics modules and recommended modules. Statistics module is mainly to complete the user data collection and uploading, and preparation for uploading data. It includes data cleaning, noise removal, data integration and conversion, data reduction and so on. Recommended module defines the interest calculation formula at first, combined with the proposed RS-TS model is designed and implemented collaborative filtering recommendation algorithm. In order to further improve the quality of recommendation, we use time decay adjustment algorithm and hits adjustment algorithm based on interest calculated RS-TS model results were then adjusted recommend achieved satisfactory recommended program set.Experiments and test run show that set-top box designed and implemented program recommendation system has a good user’s experience, satisfaction with the program recommended made a relatively high reputation, showing a very good application prospects. At the same time, this work has also certain reference value for the video class recommended research.This paper is organized as follows:Chapter1(Introduction) First describes the research background and significance, introduces the main work.,lists the organizational structure.Chapter2(Related technology and research background briefings) Introduces the current major recommendation system and recommendation algorithms. At the same time also introduces the social computing and interest computing related content.Chapter3(Overall design) General overview of the system is introduced, RS-TS model, time decay and hits adjustment algorithm. The system module division and specific features include database design are also discussed.Chapter4(Statistics module design and implementation) Introduces the statistics module design and implementation deeply, including data cleaning, noise removal, data integration and conversion, and data reduction and so on. Meanwhile, data collection module po class user survey forms, data preprocessing algorithm and other content related to the show.Chapter5(Recommended module design and implementation) Elaborated definition of interest calculation formula, combined with the proposed RS-TS model is designed and implemented collaborative filtering recommendation algorithm. Meanwhile, the time decay adjustment algorithm and traffic adjustment algorithm is introduced in detail.Chapter6(Experimental results and analysis) Describes the system’s operating environment, showing the key parts of the system run screenshot and Offers comparative analysis results. Chapter7(Conclusions and further work) Made a summary of the system, and then pointed out that the system further room for improvement and perfection directions.