摘要
针对目前金融行业普遍存在的金融产品信息过载、产品种类繁多、客户选择困难的问题,提出了一种基于知识图谱和用户画像的改进推荐方法。该算法通过知识图谱和用户画像技术分别计算企业与产品的相似度并线性加权融合,利用融合后的相似度矩阵对协同过滤算法进行改进,并设计实现了金融产品推荐系统。使用国网电商提供的真实数据集进行仿真实验,改进算法的F1值在0.6~0.7,而相同企业的原始协同过滤算法推荐效果的F1值在0.5~0.6。与原始协同过滤算法相比,改进算法有效缓解了数据稀疏性问题,提高了推荐效果。
Aiming at the problems of financial product information overload,a wide variety of products,and difficulties in customer selection,which are common in the financial industry,an improved recommendation method based on knowledge graphs and user profile was proposed.The knowledge graph and user profile technologies were used to calculate the similarity between the company and the product by linearly weighted fusion.The fused similarity matrix was used to improve the collaborative filtering algorithm,and the financial product recommendation system was designed and realized.In the simulation experiment using the real data set provided by the State Grid E-commerce Company,the F1 value of the improved algorithm was between 0.6 and 0.7,while the F1 value of the recommendation effect of the original collaborative filtering algorithm of the same company was between 0.5 and 0.6.Compared with original collaborative filtering algorithm,the improved algorithm effectively alleviated the problem of data sparsity and improved the recommendation effect.
作者
王杰
谢忠局
赵建涛
陈思安
苗祯
滕菲
WANG Jie;XIE Zhongju;ZHAO Jiantao;CHEN Si’an;MIAO Zhen;TENG Fei(School of Control and Computer Engineering,North China Electric Power University,Beijing 102206,China;State Grid Financial Technology Group Corporation Limited,Beijing 100053,China;State Grid Xiong’an Commercial Factoring Company Limited,Beijing 100053,China)
出处
《计算机应用》
CSCD
北大核心
2022年第S01期43-47,共5页
journal of Computer Applications
基金
国网电商公司科学技术项目(9100/2021⁃72001B)。
关键词
用户画像
知识图谱
金融产品
协同过滤
个性化推荐
user profile
knowledge graph
financial product
collaborative filtering
personalized recommendation
