Design Of User Portrait System Based On Analysis Of Mobile Internet Behavior

With the advent of big data age,it's vital for the development of customers to seize the old customers,attract new customers,read the consumer's preferences and dig the potential value of mining consumers.And achieving this goal requires to refine marketing for the consumer market,apply the data mining technology to make a portrait for consumers and then implement the customer segmentation.The research and development in the realization of precise marketing,improve work efficiency and reduce operating costs have played an important guiding significance and practical value.In this paper,I set the mobile Internet user's behavior as a research object and make the user portrait theory as a theoretical basis,put forward the idea about user portrait system construction and use the label method to describe the consumer's behavior and preferences characteristics.At the beginning of this paper,I expound the consumer portrait system.The consumer image system is summarized from the concrete semantics and the abstract mathematical model.Moreover,the characteristics of the user's portraits are described from the dynamic and the space-time limitations.In the end,the image of consumer is divided into static information portraits and dynamic information portraits from the perspective of modeling,and the modeling methods used in them and the data mining algorithms used are all introduced in detail.Secondly,the overall structure of the user portrait is described.Respectively,from the overall design and functional architecture of the system,as well as the business structure of the user portrait system design ideas for a detailed description.In this paper,the user profile is described in detail from the basic attribute,the business attribute,the product attribute and the channel attribute elaborated the user static information label construction method and the concrete classification according to the operation characteristic and the business experience of the mobile industry.Then,the Chinese word segmentation is realized by means of data mining method,and the eigenvector of the user's text is obtained by using the LDA clustering model to obtain the eigenvector of the user's text.Then the improved TF-IDF classification method is used to predict the different dimension information of the user,Of the Internet based on the characteristics of the property.Secondly,through the analysis of the user's online log and the matching of the web crawler data,the scoring model of the content is constructed,and the information such as the characteristics and preferences of the mobile client is highly summarized,and the user's mobile Internet preference feature tag is generated,which greatly enriches the user's dynamic information The construction of the tag.On this basis,it is proposed that the user image system uses the column database to store the image data,label metadata management,and label lifecycle management,query mechanism and update mechanism.A complete set of standardized user image system management process to build a complete Closed-loop user portrait system.Finally,the application of the user profile in the mobile marketing business,Take the traffic promotion in the marketing process and the promotion of the mobile phone software user as an example.