Research of the USIP Slope One Collaborative Filtering Recommended Algorithm Based on Hadoop

As the Internet technology grows, various kinds of information have been presented in front of us, and it is getting more and more difficult for us to retrieve the information in which we are interested. The socalled Collaborative Filtering is an efficient path to solve this problem, because it can build a model through the analysis of users' historical information, and recommend to the users the products they are potentially interested in. Slope one is a classical Collaborative Filtering recommended algorithm, however, its forecasting accuracy can be affected by the data sparsity. Therefore, this paper proposed an improved Slope One algorithm USIP (User Similarity and Item Property). Based on the user average similarity and the item attribute similarity, this algorithm calculates the prediction score of user u to the target item j through the auto-adapted adjustment of the proportion between both sides, as well as conducts parallelization on the data with the distributed Hadoop platform, so as to improve the ability of processing data. The result indicates that the improved algorithm enhances the accuracy of the prediction and the execution speed, as well as is suitable for processing the largescale data.