基于混合协同过滤的个性化推荐方法研究

针对传统协同过滤算法存在的数据稀疏性和推荐范围问题，提出一种混合协同过滤推荐方法.该方法将两种传统算法结合，并综合考虑了项目标签属性等信息.首先利用基于项目的协同过滤算法生成预测评分，并替换原始用户-项目评分矩阵中的零值.其次利用基于用户的协同过滤算法计算填充后矩阵的用户相似度，以及预测评分并产生最终推荐.最后基于MovieLens数据集实验证明，该方法能够有效提高推荐精度，扩大推荐范围.

Research on personalized recommendation method based on hybrid collaborative filtering

The traditional collaborative filtering algorithm has some problems of data sparsity and recommendation range. For these problems， this paper proposes a hybrid collaborative filtering recommendation method. The algorithm not only combines two traditional algorithms， but also comprehensively considers the item label attribute information. Firstly， the item-based collaborative filtering algorithm is used to generate a prediction score and replace the zero value in the original user-item rating matrix. Secondly， the user-based collaborative filtering algorithm is used to calculate the user similarity of the filled matrix， predict the rating and generate the final recommendation. Finally， based on the MovieLens dataset experiment， the method proposed can effectively improve the recommendation accuracy， expand the recommendation range， and has certain application value in the field of digital resource recommendation.