Research On Intelligent Recommender System Based On Cloud Computing

Nowadays we create, transmit and receive information everyday. Information is in an unprecedented expansion state in contemporary. Experts and academics have been doing some relevant research facing such vast amounts of data. Cloud computing and recommender system are two important directions of them.There are many powerful advantages of cloud computing in storing and processing massive data. Meanwhile, it provides users with storage and computing resources by service and allocates service performance flexibly according to users’ need. Usually we need to purchase computer hardware and software to get the conveniences computers bring to us. Computers would stay idle when we don’t use them or only use a small part of their resources which causes a waste of resources. It not only waste lots of money to purchase some resources we cann’t make full use of, but aslo damage the enviroment by creating such large number of computer resources in idle, especially for the eliminated old computers. However, cloud computing has changed our concept of computer. We don’t need computer hardware and software, but a kind of IT service. Therefore, we don’t need to purchase computer hardware and software, but IT service provided by full developed cloud computing. Then we can access to cloud service through any available terminal. Many leading companies in IT industry are involved in research and development of cloud computing, including IBM, Google, Microsoft, Apple and so on. There are three technical theories which are GFS, MapReduce and BigTable proposed by Google Company from the aspects of file system, programming model and relation data of cloud computing. These theories play a guiding role in the development of cloud computing. Basing on these theories, many companies and organizations realize cloud computing, and the most important implementation is Apache’s Hadoop.Recommeder system can help users find somthing they are really interested in from lots of information. The recommended method of recommender system include content-based recommendation, collaborative filtering-based recommendation and knowledge-based recommendation. Collaborative filtering-based recommendation is the most commonly used recommendation way. It finds users who share the same interest and taste with you and recommends the projects they like to you. It can reflect the collective wisdom among people in the age of Internet. There are natural interaction and impact potential between recommender system and cloud computing. The data sources of recommender system in contemporary industry are often millions, even billions of data records. It puts forward higher requirements for how the system stores these datas and how to calculate the recommender results quickly. Cloud computing technology provides the answer to these questions. In the development of future IT, cloud will become the driving force of all computing tasks and storge place.In order to achieve efficient and high-quality recommender systems in the cloud computing platform, this paper elaborates the current research situations of recommender systems and cloud computing technology and focuses on research and analysis on project-based collaborative filtering recommendation algorithm and recommendation algorithm based on association rule mining. It consults some improvement principles of Google cloud computing programming model and obtains the improved algorithm based on MapReduce item-based collaborative filtering and association rule, which makes the algorithm can be runned in Hadoop platform. It makes recommender system have strong parallel computing capabilities and scalability by allowing the algorithm to adapt to the cloud platform.