摘要
协同过滤算法是推荐系统中最为成功的技术之一,相似性计算是协同过滤算法的核心.针对传统的相似度计算方法在数据稀疏的情况下推荐不准确问题,提出了基于项目间差异信息熵的相似度计算方法,先通过差异值和共同评价数目对信息熵进行加权,再归一化处理来计算项目间的相似度.用基于项目(Item-based)相似性的协同过滤算法进行了实验验证,实验结果表明,该算法提高了个性化推荐精度.
Collaborative filtering algorithm is one of the most successful recommender system technology. The similarity calculation is the core of the collaborative filtering algorithm. In view of the poor predication quality existing in traditional similarity calculation with sparse data,we propose a similarity calculation method based on the information entropy between differences of items. First, we weight the entropy by the difference and com- mon evaluation and then normalized it to measure the similarity between items. Verified by experiments with i- tem-based collaborative filtering algorithm, the results show that it improves accuracy of personalized recom- mendation.
出处
《郑州大学学报（工学版）》
CAS
北大核心
2012年第5期118-120,共3页
Journal of Zhengzhou University（Engineering Science）
基金
国家自然科学基金资助项目(60970060)
天津市教委资助项目(20071328)
天津市科技支撑计划重点项目(09ZCKFGX00500)
天津师大博士基金项目(52LX17)
关键词
信息熵加权
相似度计算
协同过滤
个性化推荐
weighted information entropy
similarity calculation
collaborative filtering
personalized recom-mendation
