Mobile Cloud Computing Multimedia Service Technology Research

In recent years, the growth of the mobile Internet is getting explosion, the popularization of mobile equipment and the massive coverage of 4G and WIFI network makes multimedia service to become more and more convenient. Users can enjoy the multimedia service in anytime and anywhere through mobile device. Insufficient resources of mobile device itself and poor quality of service of multimedia service itself have become increasingly prominent. The emergence of mobile cloud computing effectively solves the problem of insufficient resources of mobile device. But there is a high delay problem on mobile cloud computing. Cloudlet as realizing mode of mobile cloud computing effectively solves the problem of high delay and retain the advantages of mobile cloud computing. News recommendation service as a typical multimedia service is one of the most popular services on the Internet. The inaccurate problem of the description of news content features and user interest feature exist in news recommendation service. This paper will focus on the related technology of news recommendation service and propose the corresponding solution.The main work of this paper:1. Based on Cloudlet and news recommendation service, we design a set of news recommendation system based on Cloudlet news. It aims to improve the time performance of news recommendation system by using Cloudlet.2. According to the recommendation service process which introduces Cloudlet, we design a news query and recommendation algorithm based on Bloom filter. It will reduce the news loading time and further improve the time performance of the system.3. According to the inaccurate problem of news features and user feature description which exists in traditional news personalization recommendation algorithm, we design a improved random forest news text classification algorithm based on Latent Dirichlet Allocation. We use the news features calculated by this algorithm as the basis of user interest feature modeling. And we design the recommendation solution of system based on these features. At the same time, the algorithm can balance the classification accuracy and time cost better. So it can adapt to the task of massive news text classification.This paper presents the design of complete system architecture, the recommendation process and each function module of the system. In the end, the algorithm and the recommendation solution of system designed in this paper significantly improve the performance of news recommendation system which is verified through the experiment test.