结合类别偏好信息的Item-based协同过滤算法

传统的基于项目的协同过滤算法离线计算项目相似性,提高了在线推荐速度.但该算法仍然不能解决数据稀疏性所带来的问题,计算出的项目相似性准确度较差,影响了推荐质量.针对这一问题,提出了一种结合类别偏好信息的协同过滤算法,首先为目标项目找出一组类别偏好相似的候选邻居,候选邻居与目标项目性质相近,共同评分较多;在候选邻居中搜寻最近邻,排除了与目标项目共同评分较少项目的干扰,从整体上提高了最近邻搜寻的准确性.实验结果表明,新算法的推荐质量较传统的基于项目的协同过滤算法有显著提高.

An Improved Item-based Collaborative Filtering Algorithm Combined with Class Preference Information

The traditional item-based collaborative filtering(CF) algorithm computes item-item similarity offline, so it enhances the real-time performance of recommender system. However, item-based CF algorithm still suffers from the data sparsity problem, as a result the recommendation quality is poor. To address this issue, a novel CF algorithm combined with class preference information is proposed. The proposed algorithm first finds out candidate neighbors who are similar to the target item in class preference, the candidate neighbos have similar nature and more co-ratings with the target item. Then it searches for nearest neighbors in the candidate neighbor set, which eliminates the interference of the items those have few co-ratings with the target item. The experimental results show that the recommendation quality of the new algorithm is significantly improved compared with traditional item-based CF algorithm.