基于重复消费动力学的用户网络消费模式分析  

Analysis of user consumption pattern based on dynamic of Web repeat consumption

作　　者：孙利娟[1,2] 罗刘敏[3] 杨新锋[4] Sun Lijuan;Luo Liumin;Yang Xinfeng(School of Computer Science, Wuhan University, Wuhan 430072, China;Dept, of Information Electronic, Kaifeng Institute of Education,Kaifeng Henan 475000, China;Dept, of Information Science & Engineering, Wanfang College of Science & Technology HPU, Zhengzhou 451400 , China;School of Computer & Information Engineering, Nanyang Institute of Technology, Nanyang Henan 473004, China)

机构地区：[1]武汉大学计算机学院,武汉430072 [2]开封教育学院信息电子系,河南开封475000 [3]河南理工大学万方科技学院信息科学与工程系,郑州451400 [4]南阳理工学院计算机与信息工程学院,河南南阳473004

出　　处：《计算机应用研究》2016年第9期2701-2704,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(41001292);河南省重点科技攻关计划资助项目(122102210563;132102210215)

摘　　要：为了更好地理解时效性对重复消费的影响,提出了一种包含质量和时效性的重复消费混合模型。在重复消费动力学中,用户对重复项的选取取决于先前所有相同项的时效性权重之和。在混合模型的参数求解过程中,提出了一种交互迭代的参数更新方法。由于用户的真实消费数据往往服从幂率分布,分析了幂率分布下的模型倾斜性,并给出了倾斜性的紧致下届。通过对GPlus和You Tube两个真实的数据集进行分析发现,用户的重复消费数据存在着时效性。在数据集服从幂率分布的情况下,对数据集进行拟合可以很好地对用户的重复消费进行预测。In order to better understand the effect of recency on repeat consumption, this paper proposed a hybrid model including quality and recency of items. In dynamics of repeat consumption, the selection of items for a user depended the sum of all weights from the same item. While solving the parameters of the hybrid model,this paper proposed a complementary iterative update method for parameters. Because users’ real consumption data always conformed to the power law distribution, it analyzed the tipping of the model under power law distribution, and gave a tighter lower bound for tipping. Experiments on GPlus and YouTube datasets show that,there are real recency in people5 s repeat consumption data. While the datasets conforming to the power law distribution,it can better predict people5 s repeat consumption according to fitting the datasets.

关 键 词：消费动力学 消费模式 重复消费 混合模型 时效性 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]