The Design And Implementation Of Movie Recommendation System On Smart TV

With the rapid development of smart TV technologies, smart TVs have shown the growing popularity and increasing number of users. On the other hand, with the increasing diversity of program sources for smart TVs, one smart TV can acquire media content from various sources such as Internet and PCs, providing the user more and more information, which usually leads to the so-called "information overload" phenomenon. The information overload refers to the situation that the amount of the social information exceeds that of one person or system can accept, causing the person or system lost in the massive collection of information and difficult to quickly and easily find the content they are interested. Therefore, it becomes an urgent demand for technical innovations in current smart TV development that help to screen out the contents that one user may prefer from the large amount of media information according to the user’s preferences. As one promising solution for this problem, developing effective recommendation systems on smart TV undoubtedly has broad application prospects.As presently the recommendation applications on the smart TV are still relatively rare, the work of this paper proposes one recommendation scheme for the device side of the smart TV, and correspondingly designs and implements a prototype recommendation system for the movie media on smart TVs. The system is built on the embedded platform and the development board of the smart TV, and integrates the user log collecting and recommendation modules. Many sources of data collection are embodied in the log collection module, while in the recommendation module, recommendation algorithms based on content filtering on the device side are designed and implemented. Moreover, the system employs the sqlite database and multi-threading technologies, which show well real-time properties, and can also be ported to other types of device sides. The test results of the system on experimental datasets have shown the well performance of the recommendation system, which achieves its original design goals.The paper first briefly introduces the Smart TV platform and the current development status of the personalized recommendation applications for digital contents on it. Next, the paper introduces various technical components of one recommender system on the device side of the smart TV, including the data storage, recommendation algorithm, SQLite database and XML technologies. After that, the paper presents the requirement analysis, the overall design framework and the detailed designs of each module of the recommendation system, followed by the description of implementation details of the system. Finally, on the basis of summarizing the existing work, the paper discusses the potential improvements and further work of the recommendation system on smart TVs.