摘要
针对商品推送系统的精度低和商品过载问题,文中提出一种基于用户画像特征信息的精准营销推送模型。基于用户历史行为的分析结果与评分矩阵,构建用户的兴趣模型,利用标签规则将其转化为用户的标签模型。同时根据多种混合规则调用Thrift框架接口,返回用户偏好的推送商品列表,从而实现具有优异性能与较低延迟时间的精准营销推送系统。仿真结果表明,文中的精准营销推送系统使用NDCG算法进行参数调优,具有较高的精确度和稳定性。
In allusion to the problems of low precision and overload of commodity pushing system,a precise marketing pushing model based on user′s portrait feature information is proposed.Based on the analysis results and rating matrix of user′s historical behavior,user′s interest model is constructed,which is transformed into a user′s label model by means of the label rules.The Thrift framework interface is called according to multiple hybrid rules to return the list of pushing products preferred by users,so as to realize the precise marketing pushing system with excellent performance and short delay time.The simulation results show that the precision marketing pushing system in this paper can use NDCG algorithm to optimize parameters,which has high accuracy and stability.
作者
黄位华
范欣
HUANG Weihua;FAN Xin(Institute of Science and Technology,Nanchang University,Gongqingcheng 332020,China)
出处
《现代电子技术》
2021年第6期43-46,共4页
Modern Electronics Technique
基金
江西省教育厅科技项目(161521)。
关键词
营销推送
智能分析
兴趣模型构建
标签模型
系统设计
仿真实验
marketing pushing
intelligent analysis
interest model establishment
label model
system design
simulation experiment
