A New Prediction Approach Based On Polynomial Regression For Collaborative Filtering

With the rapid development of Internet technology, the collaborative filtering (CF) approach to recommenders has recently enjoyed much interest and progress. While the term collaborative filtering (CF) has only been around for a little more than a decade, CF takes its roots from something humans have been doing for centuries-sharing opinions with others. Using collaborative filtering algorithm. Recommender systems help users fil-ter information based on similar users’preferences. Most of existing collaborative filtering make predictions using weighted average method. This paper introduces a new predic-tion approach based on an effective quadric polynomial regression model. One basic idea behind this approach is that there exist patterns among different users’preferences. And we propose a quadric polynomial regression model to characterize the inner relationships among different users’ratings. The major contribution of this approach is that it can make more accurate predictions via utilizing the exact quadric polynomial correlation indicated by Pearson Correlation Coefficient directly. The preliminary experiments on MovieLens data sets show that our approach can improve the accuracy of prediction thus make recommendations more appealing to users.In addition,base on Spearman’s rank cor-relation coefficient,we introduce a new similarity formula.and discuss the differences with Pearson Correlation Coefficient when calculating the similarity on Matlab.