USO- a new Slope One algorithm based on modified user similarity
Mingtao Sun, Hui Zhang, Shiyu Song, Kejia Wu · 2012
Recommendation algorithm based on collaborative filtering strategy has achieved great success in the field of personalized recommendation, However, based on collaborative filtering recommendation algorithm complexity is too high, over-fitting problem, the recommended accuracy requirements lead to the algorithm difficult to achieve in practical applications. Slope One algorithm greatly simplifies the complex process of collaborative filtering algorithm, dramatically reducing the recommended system and maintenance difficulty, but at the same time recommended the accuracy will decline. Recommended in order to take into account the recommendation algorithm of low complexity and high accuracy, the authors propose the improved algorithm based on the correct user similarity Slope One - USO algorithm The inspection results show that the algorithm can effectively improve the accuracy of the recommendation of the slope-one algorithm.