Research On Collaborative Filtering Recommendation Algorithm Based On Improved Weighted Slope On

With the development of Internet and e-commerce,more and more people rely on a series of online services.However,the number of commodities is growing increasingly.In order to help people to find their favorite products among large number of items,recommendation systems are developed.Recommendation systems can provide the items which users interested in.The core issues of the recommendation systems are making efforts to recommend matching products to users,such products should be personalized,affordable and with high degree of matching ratings.Now,the existing recommendation technologies provide solutions for different areas(such as e-commerce sites,Movie,video website,Internet Music Radio and so on).The real site instances include Amazon,Netfix,CDNow,Dangdang,Douban,etc.Personalized recommendation systems have good prospects,but these systems are also facing many problems.With the further research,these questions have been resolved to a certain extent,and have created satisfactory recommendations in movie,music,and some other fields.Collaborative filtering is one of the most widely used recommendation techniques,but data sparsity probllems resolved to the low recommendation accuracy and efficiency.Slope one is a simple collaborative filtering algorithm,it is based on linear regression model and utilizes rating deviations to predict ungraded items.However,Slope one algorithm is not better than traditional collaborative filtering algorithm when data is sparse,so this algorithm cannot resolve the data sparse problem.Many improved Slope one algorithm have been proposed to obtain higher accuracy in recommendation.The current existing studies are all based on the basic Slope one algorithm,and make prediction on the nearest items.Since the user’s nearest neighbors are different,they have various preferences to the items.In this paper,an improved Slope one algorithm based on Mahout is proposed,this algorithm is ground on user-based collaborative filtering algorithm,Specifically,the algorithm firstly calculates the user’s nearest neighbors by the similarity between users,and then finds the appropriate number of neighbors based on user-based collaborative filtering.In the end,weighted Slope one algorithm is used within the user’s nearest neighbors to predict empty ratings and to giye recommendation.The main work of this thesis is as follows,this paper focused on collaborative filtering algorithm,and based on the analysis and comparison of Slope one algorithm.Then,through the contrast experiment between the basic algorithm and improved algorithm,the improved algorithm is proved to be much better.