摘要
针对用户评分数据极端稀疏情况下传统个性化推荐算法的不足,提出基于平均差异度的个性化推荐算法,该算法通过计算用户对项目评分之间的平均差异度来预测用户对未评分项目的评分,从而产生高质量的推荐。实验结果表明,该算法可以有效地提高数字图书馆个性化推荐系统的可扩展性及推荐准确度。
Against the extremely sparscl user rating data with traditional personalized recommendation algorithm, an personalized recommendation algorithm is proposed based on the average difference degree, by calculating the average difference between the item ratings for users who rate both, so as to produce high-quality recommend result. The experimental results demonstrate that the algorithm can effectively improve scalability and accuracy of the digital library of personalized recommendation system.
出处
《图书情报工作》
CSSCI
北大核心
2009年第11期119-122,共4页
Library and Information Service
关键词
数字图书馆
个性化推荐
平均差异度
digital library
personalized recommendation
average difference degree
