Research On Collaborative Filtering Algorithm Based On Forgetting Function And User

With the development of the Internet,as in music,shopping,reading,film etc,users increasingly depend on the network to bring convenience.The increase in the number of the users and the new products cause the system need store a large amount of data.In the "data overload",if we can obtain valuable information from the data,then we can provide users with better service.Therefore,in the e-commerce,social networking,video sites and other platforms,personalized recommendation technology research has been widely used.At present,the most widely used personalized recommendation technology is collaborative filtering recommendation technology,which analyzes user's historical behavior records to get the product that user interested in and recommend to the user.This paper take user's interest change phenomenon exists into account.However,collaborative filtering can not based on the user's actual process of forgetting to give the right weight to the project,so the recommendation is not accurate.In this paper,the nonlinear Ebbinghus forgetting algorithm which is obtained by collaborative filtering is used to describe the user's interest migration.The collaborative filtering based on the forgetting function and user's add the forgetting factor weight in calculating user similarity to reflect the change of user's interest.We use the improved algorithm to test and evaluate the MovieLens and the EachMovie in the movie data set.The experimental results indicate that the improved algorithm can improve the accuracy of recommendation in two different movie related data sets.Therefore,after fitting the forgetting curve,the forgetting function can describe the user's interest migration and this provide the improved direction of reference for the website recommendation.