Abstract: Slope One algorithm is a widely used item-based collaborative filtering recommendation algorithm. But the algorithm does not consider the impact of the item type factors in the process of calculation of the item similarity, leading to type irrelevant items involved in the similarity calculation which influences the effectiveness of the recommendation, and the prediction accuracy is not high in the case of sparse data. In order to solve these problems, this paper considers using the item relevance to measure the relationships between items, and proposes a weighted Slope One algorithm incorporating the relevance of item. The algorithm synthetically calculates the similarity relationship between the items from two aspects: the type of the item itself and the scoring data of the item. On this basis, the item's culling strategy is added, which achieves the goal of stable score difference and getting more intensive item scoring matrix. Finally, the proposed algorithm is applied in the MovieLens data set. The experimental results show that the proposed algorithm not only improves the accuracy of recommendation, but also alleviates the data sparseness problem effectively.

Key words: item relevance, weighted Slope One, collaborative filtering recommendation, data sparseness

摘要： 在基于项目的协同过滤推荐算法中，Slope One算法的应用较为广泛。但该算法在对项目相似度计算过程中没有考虑项目类型因素的影响，导致类型不相关的项目参与到相似度计算中而影响推荐效果，而且该算法在数据稀疏的情况下预测准确度不高。针对上述不足，考虑使用项目相关性来考量项目间的关系，提出了融入项目相关性的加权Slope One算法，该算法从项目自身的所属类型和项目的评分数据两方面来综合计算项目间的相似度关系，并在此基础上增加了项目筛选策略，达到了稳定评分差、获取局部较为密集的项目评分矩阵的目的。最后将所提算法应用在MovieLens数据集上与其他相近算法进行对比实验，实验结果显示所提算法在明显提高推荐准确度的同时也有效缓解了数据稀疏问题。

关键词: 项目相关性, 加权Slope One, 协同过滤推荐, 数据稀疏