Research On Recommendation Algorithm Of WFSLIM Based On Improved Fuzzy Clustering

With the development of Internet technology,people have entered from the PC era to the smart phone as the representative of the mobile Internet networking era,the amount of human made much data generated every day.It is also known that the 21 st is the era of data(DT).Mobile Internet to produce huge amounts of data at the same time also greatly promoted the microblogging,Twitter,Taobao and other social and business network of hot development.According to CNNIC reported that as of the end of 2016,the total number of Internet users in China has reached 710 million.Such a large Internet users accompanied by a huge amount of data.How to use the data generated by the user to carry out efficient and economical analysis of personalized has become an important research problem.At present,regardless of microblogging and other social networks or Taobao and other e-commerce site of the existing recommended system effects are often not so satisfactory.The traditional recommendation system is mainly based on the recommendation of the score,while the social and electric business recommendation system is mainly focused on the Top-N recommendation.A good automated recommendation system can greatly improve the user's activity in the platform revenue.In this paper,by introducing the shortcomings of various traditional recommendation algorithms,this paper proposes to use the improved WDFCM clustering algorithm to cluster the users,and finally consider the influence of the correlation coefficient between the users and the project on the forecast:(1)Many of the recommendation systems about commercial and social platforms primarily use model-based collaborative filtering algorithms.This algorithm often has data sparsity,poor system scalability,cold start and other issues.The recommendation system based on SVD simply makes a mathematical matrix decomposition of the scoring matrix,which is only a direct relationship between the user and the commodity.And did not fully take into account the impact between the user and the project,the matrix decomposition of the user factor matrix is not too much to consider their specific systemin the specific significance.(2)FCM fuzzy clustering algorithm in dealing with some specific scenes,the experimental results of the great effect depends on the parameters and the choice of the center point,this paper takes into account the fuzzy clustering feature distance and weight on the accuracy of clustering,WDFCM algorithm,through experiments to verify the algorithm for some data sets,clustering accuracy has been significantly improved and reduced the number of iterations of the cluster.(3)In this paper,the semantic semantic model of the proposed algorithm based on the model is extended,and the correlation coefficient between the users and the project is taken into account in the expansion model.In view of the sparse and cold start problem of the traditional recommendation algorithm,this paper proposes to make the WDFCM Clustering,taking into account the impact of the correlation coefficient between the user and the user set,the user and the user,the project and the project on the scoring matrix,and then recommend the improved WFSLIM recommendation algorithm.So as to reduce the scarcity of the scoring matrix and improve the accuracy of the model recommended,under certain conditions,reduce the running time of the algorithm.(4)With the rapid increase of users and data of e-commerce websites,the traditional recommendation algorithm platform has been obviously deficient in dealing with massive data.At the end of this paper,the design and development of recommendation system algorithm is realized on Hadoop large data platform.