Research On Spark Book Recommendation Technology Based On Campus Resource Cloud

With the advancing and deepening of information construction in colleges and universi-ties,the construction of campus cloud platform has been drawn much attention by numerous universities.Not only can it meet and ensure the needs of schools in all aspects,but also pro-vide an efficient and reliable computing storage platform for campus big data analysis.The research of this subject depends on the campus resources cloud platform,and also gains a strong support from information infrastructure.Meanwhile,the data are accumulating due to the extensive applications of different management information systems in business.Particu-larly,the library management system has accumulated a large number of historical data for books circulation and the data in the system increases continuously as time goes on.However,there is a great deal of valuable potential information behind these data.To make full use of Library Circulation Data and improve the information experience of teachers and students,this thesis has a further analysis and research so that teachers and stu-dents can get personalized book recommendation service.This thesis,firstly,designs in com-puting,storage resource and platform function for campus resources cloud platform.Then it takes cloud platform as test and operation platform for book recommendation,and built Spark cluster on it.This thesis takes HDFS as the storage system and Spark as the computing plat-form to do some studies on the technology of book recommendation.Aiming at the problem of data loss and data form,this thesis does pretreatment on the raw data,and constructs the user-book score matrix.In addition,the ALS matrix decomposition collaborative filtering al-gorithm is employed to deal with the problem of data sparsity,and the K-Means clustering algorithm is integrated into ALS matrix decomposition algorithm to solve the user cold-start problem.Besides,regarding the problem of attribute weights and initial values of K-Means algorithm,the weighted Euclidean distance and the maximum and minimum algorithm are used to optimize it.Finally,the algorithm is implemented on Spark,and the experiments are carried out to verify the design.It implements personalized book recommendation for differ-ent users.According to the experiments,this thesis determined the optimal parameters of ALS ma-trix decomposition algorithm,and demonstrated that the proposed hybrid algorithm can solve the problem of data sparsity and cold-start.Furthermore,the optimization of K-Means algo-rithm improves the clustering effect,and the integrating of clustering algorithm improves the prediction accuracy and calculation speed.Finally,the advantage of Spark cluster is verified by the speed-up of parallel computing in Spark platform.