Abstract: Compared with the collaborative filtering algorithm, the Slope One algorithm is faster and easy to implement, however, the algorithm does not consider the recommended level between items. Therefore, this paper proposes a Slope One recommendation algorithm based on weighted bipartite graph. The weighted bipartite graph algorithm calculates the recommended level between the items and uses the recommended level to weight the prediction of the Slope One. Since the recommended level between the items calculated by the bipartite graph recommendation algorithm is asymmetric, the results are more diversified.We test the proposed method by 5-fold cross-validation on the MovieLens with four metrics: ranking score, Recall, Precision and diversification, and the experimental results show that the improved algorithm improves the recommendation accuracy and improves the recommendation diversity.