融合用户特征和流行度归一化的金融产品混合推荐算法  

Hybrid recommendation algorithm for financial products that integrates user characteristics and popularity normalization

作　　者：王珺 胡荣婕 WANG Jun;HU Rongjie(School of Mathematics&Statistics,Changchun University of Technology,Changchun 130012,China)

机构地区：[1]长春工业大学数学与统计学院,吉林长春130012

出　　处：《长春工业大学学报》2023年第2期117-122,共6页Journal of Changchun University of Technology

基　　金：国家自然科学基金项目(11871244);吉林省科技厅自然科学基金资助项目(20200201273JC)。

摘　　要：混合推荐通过组合不同推荐算法来弥补各自推荐技术的弱点。在传统金融产品推荐算法的基础上,提出融合用户特征和流行度归一化的金融产品混合推荐算法(CPCF)。在Santander银行客户数据上进行实验,将基于人口统计学的推荐算法与协同过滤推荐算法进行组合,在此基础上,引入流行度权重因子,将流行度归一化处理。实证结果表明,在Precision和Recall上均有提升,适合用户信息容易采集且用户数量级远超过项目数量级的金融理财产品推荐场景。Hybrid recommendation compensates for the weaknesses of their respective recommendation techniques by combining different recommendation algorithms.On the basis of the traditional financial product recommendation algorithm,a financial product hybrid recommendation algorithm(CPCF)that integrates user characteristics and popularity normalization is proposed.In the experiment on Santander Bank′s customer data,the demographic-based recommendation algorithm and the collaborative filtering recommendation algorithm are combined,and on this basis,the popularity weight factor is introduced to normalize the popularity.The empirical results show that both precision and recall have been improved,which is suitable for financial wealth management product recommendation scenarios where user information is easy to collect and the order of users far exceeds the order of magnitude of the project.

关 键 词：协同过滤 人口统计特征 流行度归一化 混合推荐 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]