Research On CPS Data Processing Algorithm Based On Collaborative Filtering

CPS technology has an impact on the information interaction between the physical world and the information world.Under this influence,the data resources in the network space are growing and accumulating at an unprecedented rate.People's demand for realtime storage,processing and analysis of information has been increasing,making the information and physical fusion system based on big data become a new trend.In order to meet the purpose and needs of different huge data users,various heterogeneous data fusion and processing recommendation algorithms have been put forward and applied,among which the most famous is collaborative filtering recommendation algorithm.According to past research,although collaborative filtering recommendation algorithm has been successfully applied in the field of recommender system,there still exist some problems such as scalability,data sparsity and cold start.The quality of data processing will be reduced by the afore-mentioned problems.Therefore,this study uses normal similarity measure to correct the error of the collaborative filtering algorithm,and Hadoop in the cloud environment,and compare the measurements with single single execution time in 3,6 and 9 nodes,accelerate the analysis and improvement of collaborative filtering algorithm and performance ratio.The main contents and main conclusions of the study are as follows:Firstly,based on user based collaborative filtering algorithm,the BP neural network algorithm is applied to improve the similarity measure of normal recovery of collaborative filtering algorithm,so as to improve the calculation error of collaborative filtering recommendation algorithm and improve the time acceleration ratio of CPS heterogeneous data.Secondly,build the experimental environment and use MapReduce on the Hadoop platform to compare and analyze the different hosts with different experimental data.Compare the results of Jaccard,Pearson and the normal algorithm.The algorithm is validated in the recommended results and time acceleration ratio and its digital processing efficiency.Finally,the experimental results show that when the number of neighbors is increased(K),the execution time is proportional to the growth.When the number of nodes is 6 and 9,the running time is greatly improved,and the efficiency of the improved collaborative filtering algorithm is significantly improved.In the 6 node,the improved cooperative processing project increased 1 times,in 9 nodes,increased to more than 4 times,the highest speedup in the number of neighbors K equal to 6 of the maximum speedup of 4.18 times.