基于项目内容和评分的时间加权协作过滤算法

Time-weighted collaborative filtering algorithm based on item content and rating

文中围绕传统的协作过滤推荐算法存在的局限性展开研究,提出了一种基于内容和评分的时间加权协作过滤算法。首先,计算用户已评分项目的时间权重,在此基础上,分别计算项目间基于内容和基于评分的时间加权相似度的值;然后,将二者相结合,计算用户间的相似度,形成兴趣更加接近的邻居集,进而进行高质量的推荐。实验结果表明,该算法不仅提高了推荐精度,降低了数据的稀疏性,而且算法的扩展性也得到了有效改善。

This paper has studied the limitations of the traditional collaborative filtering （CF） recommendation algorithm and proposed a time-weighted collaborative filtering algorithm based on item content and rating. First, we calculated the time weight of the item rated by users, then the time-weighed similarity of the item based on content and rating respectively. Finally, we combined both of them and calculated the similarity between users so that a closer neighbor set and high quality recommendations were obtained. The results show that the algorithm has improved the accuracy and scalability of recommended systems, and decreased the data sparsity.