Design And Implementation Of User Profile System Based On Hadoop

In this era of rapid development of the Internet,network and information technology are developing with the trend of the times,and the data of Internet users has also exploded.All kinds of daily behaviors of users are recorded through the Internet.For all Internet enterprises,if they want to stand out from the fierce competition,they need to focus on users and understand the real intentions of users through these huge user behavior data.User portrait is the description of user's data,information and labeling.Enterprises can abstract the user's business panorama through user profile,so as to carry out refined and personalized marketing for each user group.But at the same time,many enterprises have multiple business systems,and due to the incomplete information between each business system,it is unable to build a unified and clear user profile.Based on the above background,this paper designs a user profile system with high flexibility and expansibility,which integrates a variety of business data sources and establishes an enterprise system of user tags.The system covers a variety of platform channels and multi business system user source data collection,and realizes the construction of user profile based on multi-source data fusion.At the same time,according to the design idea of lambda architecture,the system framework of this paper integrates off-line computing and real-time computing,taking into account the processing capacity and timeliness of massive data,ensuring the final consistency of real-time and off-line data.The front-end of the system uses Vue.js framework and Echarts open source chart library technology to visualize the user's profile.The back-end uses Spring Boot which is a lightweight open source framework,and builds data warehouse based on Hadoop which is a distributed basic framework to realize distributed data storage,calculation and query.At last,this paper proposes a user centered data model-user event model,which can record user's behavior in detail and abstract user's behavior through event and user.And this system uses the multi-dimensional analysis method to analyze the user behavior to meet the analysis needs of each department of the enterprise.After testing and running,the system realizes the user behavior data collection and unified data management of mobile,PC,small programs and other platforms,can accurately generate user labels and user group portraits,and can provide a variety of analysis methods for user analysis,and build an open and free data management and analysis platform for accurate marketing and personalized recommendation strategies of enterprises.