An Improved Collaborative Recommendation Algorithm Research Based Optimized User Similarity

With the rapid development of the big data,people gradually sank into the ocean of information.The information overload problem has become a serious negative problems in the big data era.To ease the information overload problem,the personalized recommendation technology arises at the historic moment.And the Collaborative recommendation technology is one of the most successful recommendation technology.There are lots of issues existed in traditional collaborative filtering recommendation algorithm such as data scarcities,cold boot,recommendation accuracy and timeliness.To resolve these problems,a large number of domestic and foreign scholars put up a variety of solutions,to a certain degree,which achieved relatively desired results.But there are still few proposed solutions aimed at the Rating Scale Differences Problem(RSDP)in the Collaborative filtering algorithm.The paper carried on the thorough analysis and research for the recommended accuracy of recommendation system.Combined with the solution to RSDP,we designed an improved collaborative recommendation algorithm based optimized user similarity in the paper.Firstly,the paper summarized the current situation of the personalized recommendation technology at home and abroad,analyzed and compared the mainly recommendation technology.Secondly,by analyzing the core algorithm of the traditional collaborative recommendation technology,named user similarity calculation method,we find that the neglected issues of rating scale difference for all projects in the traditional collaborative filtering algorithm existed while calculating the similarity between users within the project set.Despite the Adjusted Cosine Similarity Algorithm and the Pearson Similarity Algorithm have improved the issue,it is still existed that users have the problem of single rating scale difference of the project.When the users have significant differences for the score vectors on a common set,they may have chance to get similar resultant vector results.Thirdly,based on the detailed analysis on the RSDP of the Cosine Similarity Algorithm on the multidimensional space score vector,and the problem of the Adjusted Similarity Algorithm and the Pearson Similarity Algorithm on the multidimensional space scale difference vector,we find a solution to balance the difference.That is,we added a balance factor that is based on a single project scale difference calculation to the traditional User Similarity Algorithm,which result on proposing an improved collaborative recommendation algorithm based optimized user similarity.At last,we obtained the most appropriate balance factor threshold by experiment.And then,we designed a series of reasonable experiments to validate the effectiveness of the proposed algorithm based on the threshold.Experimental results show that the proposed improved collaborative filtering algorithm based on user similarity can improve the accuracy of user similarity,and get better recommendation results.