The Design And Implementation Of Recommendation System For Smart Devices Based On Cloud Computing

In recent years, the popularity of mobile devices and the development of Mobile Internet bring new user habits and consumption patterns. A various kinds of mobile applications has been strongly recommended by more and more users. The mobile terminal has become the user’s personal assistant, and we can expect that the mobility and intelligent have become the trend of future Internet. More and more applications that appeared and the rapid development of network make people’s life more convenient and colorful. But at the same time, information overload became a problem. Although classified directory and search as public service have greatly enhanced the efficiency of obtaining information, but the unicity distract the user and reduces their experience.With the development of technology, recommendation system has become the most important tool to solve the problem of informations overload. It can intelligently and actively help user quickly and efficiently find valuable content in massive data based on user interest characteristics and behavior patterns. The most mature algorithm used in current recommendation system is Collaborative filtering algorithm. But it encountered many problems when met the challenge of massive data and limited processing ability in mobile devices such as the cold start, data sparsity, scalability etc. This paper aims at setting up the recommend system in the cloud platform for intelligent terminal equipment, helping users quickly find interesting diet and exercise instead of browsing a large number of irrelevant information with the assistance of algorithm and big data.In order to realize the personalized recommendation on diet and exercise, this paper constructs a collaborative filtering recommendation engine based on cloud computing platform. Users can get targeted advice information produced by the background server cluster which deployed in the cloud platform. The information displayed on users’ mobile terminal when they enter the application or refresh the page. The following are some summary and results in the process of designing and implementing the system:1. In order to support data expansion and the extension of algorithm, this paper built a recommendation engine based on Hadoop which is an open source platform, learned HDFS and Map Reduce which expanded in Hadoop, and studied the Mahout algorithm library which can solve the problem of data mining using cloud computing.2. In order to achieve the accuracy of recommendation, a various recommendation engine are combined used for the recommendation. The similarity is calculated by different algorithms, and the use of their respective advantages can achieve good effect. The recommendation engine can improve the recommendation accuracy and user satisfaction by the use of similarity information and the request parameters.3. In order to return the Top-N to the interface, online and off-line calculation are separated and combined for the quick calculation of result according to the real-time of recommendation. Interface is the link between an IOS device and the recommendation engine. The recommendation system(server) and the client(IOS devices) the paper designed transfer data with each other through the calling of RPC Server database by XMLRPC.The goal of this paper is to build a recommendation engine platform for the intelligent terminal. A recommendation engine platform architecture model which based on Hadoop and Mahout is proposed. The whole system is observed and assessment by running multiple recommendation engine on Hadoop and selecting Movie Lens and three indexes of which as the experimental data set. The feasibility of the system is proved by the accurate recommendation results in the test experiments.