Design And Achievement Of Product Personalized Recommendation System Based On User Portrait

China is a large country of Internet shopping.In recent years,with the rapid development of online shopping,some business platforms will recommend a lot of commodity information to customers by using big data.However,it is undeniable that there is still a lot of commodity information obtained through retrieval.In the face of a wide range of commodities,many consumers have no way to start,do not know how to choose,and sometimes they are not interested in recommending.In view of this situation,it not only does not bring convenience to customers,but also brings some troubles to consumers.Jingdong Mall has accumulated a lot of customers during its operation and stored every consumption record of customers.How to make full use of these data,analyze customers' interests,consumption ability,purchase channels and daily behavior,and obtain the personalized personal portrait of users in the system,so as to realize the accurate marketing of customers has become the main direction of research at present.After the design of the big data platform is formed,the user portrait can be constructed.Firstly,to build the user portrait,we should determine the label and label weight,then obtain the label data,and finally build the user portrait according to the information and label rules.The construction of user portrait model includes:functional architecture design,user portrait engine design and user portrait model.After constructing the user model,we can quickly lock in the personalized needs of consumers.Through experiments and comparative analysis on public data sets,we propose magfm algorithm,which can classify user characteristics and further recommend products to users.Secondly,through the system demand analysis,the function division of the system is further determined: data collection and storage,user portrait engine,precision marketing function,system configuration function and so on.Thirdly,the performance of the system needs to be more clear.The idea of hierarchical architecture mode is adopted here.The system is divided into different levels.Each level has independent business.Of course,different or adjacent levels can interact.The data from other application systems are stored in the database,and the interface between the system and other application systems is very important.Finally,based on the system design blueprint,the paper codes and realizes all the functions of the system.The development language is Java,the system data management is mysql,and the SSM development architecture of the system is configured and built.After the system software test,it is found that the development of the recommendation system meets the expected goal.The tested system is put into use in Jingdong Mall.Through the design of this system,it can not only help lock customers faster and more accurately,but also achieve the construction of user portrait and accurately recommend products of interest to customers,which is of great significance for the development of Jingdong Mall.