An improved slope one algorithm based on tag frequency

Technology of Collaborative filtering algorithm recommends items to the target user with information of other users who have similar tastes with the target user. The basic Slope One collaborative filtering algorithm simply uses the linear regression model to predict the ratings of items. Based on the Slope One scheme, an improved algorithm considering the frequency of items´ tags information is proposed in this paper. Firstly, the tag frequency of rated items is chosen to represent the vector of the target user´s preference. After that, the similarity between rated items and unrated one is calculated and according to it, the unrated items´ rating can be predicted. Experiments on MovieLens dataset show that the proposed approach gives better prediction accuracy.