Design And Implementation Of Recommendation System Based On Social Network For Mobile Terminal

With the development of Internet technology, we have entered an era of data explosion, users highly demand for personalized service, they no longer meet in a unified user experience, but hope that the system is able to provide the recommended content and targeted services based on their interests preferences. Recommended system came into being in this context.After market research, a domestic telecom operator have a large number of mobile applications,but these applications come from different developers, and have high marketing costs, and user participation is relatively high, they urgently need a recommendation system to connect applications and users to integrate all applications, reduce marketing costs, thus reducing the degree of user involvement.Based on the above scenario, the paper designs and implements a recommendation system for mobile terminals based on social network in order to provide an independent third-party application, on the one hand, to make the application promote to the target users as soon as possible, on the other hand, to make users find themselves interested applications quickly. The system realizes user management function to maintain the users social relations information, and provides simple ratings plugin to make third-party application access conveniently and make user score, combined with the user’s social relations information and scoring information to improve the traditional collaborative filtering algorithms and solve the sparse data and cold start problems. In addition, in order to increase the value of the system, the system provides file management functions based on the network storage to achieve file upload, download, share, synchronization and other functions.This is an innovative internship activity of IBM’s "Blue Road" project, and the project plan is reviewed by IBM experts, there is a certain application prospect. The main works in this paper are as follows:· This paper discusses the background and significance of the project, a basic overview of the social networks and collaborative filtering recommendation algorithm research status at home and abroad.· This paper describes the implementation of the collaborative filtering algorithm, and analyze the advantages and disadvantages, and propose the solution for sparse data and cold start problems for the traditional collaborative filtering recommendation algorithm, that is, combined with the social information to improve the algorithm, the user relationship is modeled as a weighted undirected graph, and sparse the user diagram, and then cluster the sparse diagram to obtain Top-K nearest neighbors, and then calculate the rating matrix based on the ratings of Top-K nearest neighbor to generate Top-N recommendation.· Based on the previously mentioned algorithm and structure, this paper designs and realizes the recommendation running on Android platform.