Research And Design Of Recommender System Based On Cloud Computing

In recent years,Internet develops rapidly and changes in each passing day.The rapid popularization and powerful function of the computer and mobile phone makes it possible that people can find out domestic and foreign information,view the news,follow with entertainment fashion,get financial advice,watch the humor jokes and buy goods,which has brought great convenience.At the same time,the amount of data also grows explosively.The amount of data generated on the Internet every day has reached the level of PB.In the face of huge amounts of information,it is extremely difficult for users to find the content they like and get useful information quickly.To deal with these problems,the recommender system is born,which can make personalized recommendations through recording and analyzing the user's daily Internet behavior data.We need to conduct two aspects of research to realize this recommender system.Firstly,we need a good recommendation algorithm to improve the accuracy of recommendation.Secondly,we need to deal with data quickly to reduce the running time of the algorithm.At present,collaborative filtering algorithm is widely used in engineering,in addition to the algorithm itself needs to be optimized,the era of big data on the efficiency of the algorithm is also proposed.There are many kinds of calculation methods of similarity in collaborative filtering algorithm,but in most practical application,only one algorithm is chosen to deal with the problem,which reduces diversity of coverage for users' preferences and makes the accuracy of the algorithm greatly reduced.At the same time,most of the research focus on the accuracy of the recommender system,while ignoring some other indicators.Therefore,it is also necessary to study how to evaluate the recommender system completely.In this paper,Hadoop cloud computing platform is used to research the technology of Hadoop cloud computing platform,the User CF and Item CF of collaborative filtering recommendation algorithm are both researched.The proposed algorithm is deployed on the Hadoop cloud computing platform to realize the parallel computation and improve the efficiency of the algorithm.Then,the evaluation system of recommender system is established,and the overall performance of the recommender system is evaluated through precision,recall,MAE and speedup.Finally,two experiments are accomplished.The first one is establishing a movie recommender system based on the cloud computing Item CF andanalyzing its performance through user' rating data on the internet.And the second one,in a job recommender system,many algorithms are utilized and compared.Results of the two experiment have proved the high performance of the cloud computing platform,and achieved the expected effect.Recommender system based on cloud computing in this paper has improved the recommendation accuracy through the combination of algorithm and performance of the recommender system through utilizing Hadoop cloud computing platform.These are of great significance in practical engineering.