基于熵优化近邻选择的协同过滤推荐算法

发布时间：2018-05-08 15:07

本文选题：协同过滤 + 近邻选择　；参考：《计算机应用研究》2017年09期

【摘要】：协同过滤推荐系统的近邻选择环节中不仅没有考虑目标项目对用户间相似性计算的影响,而且也未考虑邻居用户对目标用户的推荐贡献能力,导致既降低了相似性计算的准确性,也提高了近邻集合中伪近邻的比例。针对这些问题,提出了一种基于熵优化近邻选择的协同过滤推荐算法。算法使用巴氏系数计算项目间相似性,并以此为权重加权计算用户间相似性。引入熵描述用户评分分布特性,根据评分分布差异性衡量邻居用户的推荐贡献能力。最后,利用双重准则共同计算推荐权重,并构建近邻集合。实验结果表明,该算法能够在不牺牲时间复杂度的条件下准确地选取近邻集合,提升推荐准确度。
[Abstract]:In the nearest neighbor selection of collaborative filtering recommendation system, not only the influence of target items on the similarity calculation among users is not considered, but also the contribution ability of neighbor users to target users is not considered. It not only reduces the accuracy of similarity calculation, but also increases the proportion of pseudo nearest neighbor in nearest neighbor set. To solve these problems, a collaborative filtering recommendation algorithm based on entropy optimization for nearest neighbor selection is proposed. The algorithm uses pasteurian coefficient to calculate the similarity between items, and uses it as a weight to calculate the similarity between users. Entropy is introduced to describe the characteristics of user rating distribution, and the recommendation contribution ability of neighbor users is evaluated according to the difference of score distribution. Finally, the recommended weights are calculated by using double criteria and the nearest neighbor set is constructed. The experimental results show that the proposed algorithm can select the nearest neighbor set accurately without sacrificing the time complexity and improve the recommendation accuracy.
【作者单位】：国家数字交换系统工程技术研究中心;
【基金】：国家自然科学基金创新群体项目(61521003);国家自然科学基金资助项目(61171108) 国家科技支撑计划资助项目(2014BAH30B01) 国家“973”计划资助项目(2012CB315901,2012CB315905)
【分类号】：TP391.3

【相似文献】