基于选择性预测策略的协同过滤推荐算法

Collaborative Filtering Recommendation Algorithm Based on Selective Prediction Strategy

针对传统协同过滤推荐算法在用户评分数据极端稀疏情况下无法取得令人满意的推荐质量问题,结合User-based和Item-based协同过滤算法思想,提出了一种基于选择性预测策略的协同过滤推荐算法,算法利用高相似度阈值来计算用户相似性和项目相似性,并通过形成用户最近邻居集和项目最近邻居集来预测填充评分矩阵。基于Movielens数据集的实验表明,改进的算法有效改善了传统协同过滤推荐算法的数据稀疏性和扩展性问题,明显提高了系统的推荐质量。

The user rating data in traditional collaborative filtering recommendation algorithm are extremely sparse , which results in poor recommendation quality .A recommendation algorithm based on selective prediction strategy was proposed .The user-based recommendation algorithm was combined with item -based recommendation algorithm .The user similarity and the item similarity were calculated by high similarity threshold and the user-item matrix was evaluated by finding the neighbors of users and items.The experimental results based on MovieLens data set show that the improved algorithm could solve the problem of data sparsity and scalability , and it could improve the accuracy of system recommendation significantly .