融合用户兴趣分布变化和特征差异的协同过滤推荐算法
|
摘要
针对传统协同过滤算法没有考虑由时间引起的用户兴趣分布变化、致使其推荐精度不高的问题,提出了融合用户兴趣分布变化和特征差异的协同过滤推荐算法。采用窗方法估计用户在整个项目空间上的兴趣分布,设计时间遗忘曲线因子用以确定用户兴趣分布变化函数,最后结合兴趣分布变化相对熵和用户特征差异计算用户相似程度并进行项目推荐。实验结果表明,该算法能够有效追踪用户对项目兴趣变化,提高了数据稀疏情况下的推荐精度。
Aiming at the problem that traditional collaborative filtering recommendation algorithm failed to consider user interest change of distribution to cause poor recommending precision, a collaborative filtering recommendation algorithm combined with user interest change of distribution and characteristic difference is proposed in this paper. Window estimation method is applied to get user interest distribution in total item space, and the factor of time forgetting curve is designed to define the function of user interest change of distribution. Finally, by combining Kullback-Leibler divergence of user interest change of distribution and characteristic difference, user similarity is calculated to finish the item recommendation. Experimental result shows that the algorithm can effectively trace the interest change of distribution and raise the recommendation precision.
Aiming at the problem that traditional collaborative filtering recommendation algorithm failed to consider user interest change of distribution to cause poor recommending precision, a collaborative filtering recommendation algorithm combined with user interest change of distribution and characteristic difference is proposed in this paper. Window estimation method is applied to get user interest distribution in total item space, and the factor of time forgetting curve is designed to define the function of user interest change of distribution. Finally, by combining Kullback-Leibler divergence of user interest change of distribution and characteristic difference, user similarity is calculated to finish the item recommendation. Experimental result shows that the algorithm can effectively trace the interest change of distribution and raise the recommendation precision.
引文
[1]HERLOCKER J.Clustering items for col aborative filtering[C]//Proceedings of ACM SIGIR Workshop on Recommender Systems,New York,USA,ACM Press,1999:1-4
[2]郝雅娴,孙艳蕊.K-近邻矩阵分解推荐系统算法[J],小型微型计算机系统,2018.4(24):755-758
[3]Deshpande M,Karypis G.Item-Based top-N recommendation algorithms.ACM Trans on InformatioSystems,2004.22(1):147-177
[4]Degemmis M,Lops P,Semeraro G.A contentcol aborative recommender that exploits wordnet-based user profiles for neighborhood formation.Journal of User Modeling and User-Adapted Interaction,2007.17(3):217-255
[5]王鹏,王晶晶,俞能海.基于核方法的User-Based协同过滤推荐算法.计算机研究与发展,2013.50(7):1444-1451
[2]郝雅娴,孙艳蕊.K-近邻矩阵分解推荐系统算法[J],小型微型计算机系统,2018.4(24):755-758
[3]Deshpande M,Karypis G.Item-Based top-N recommendation algorithms.ACM Trans on InformatioSystems,2004.22(1):147-177
[4]Degemmis M,Lops P,Semeraro G.A contentcol aborative recommender that exploits wordnet-based user profiles for neighborhood formation.Journal of User Modeling and User-Adapted Interaction,2007.17(3):217-255
[5]王鹏,王晶晶,俞能海.基于核方法的User-Based协同过滤推荐算法.计算机研究与发展,2013.50(7):1444-1451