Design And Implementation Of User Portrait System For Internet Access Service Industry

The Internet access service industry,represented by the provision of computers and other Internet access environments,has played an important role in solving the problems of Internet access for the floating population and enriching people's entertainment and leisure activities by providing public Internet service venues.However,with the widespread popularity of the mobile Internet,the Internet access service industry has been greatly affected,and the entire industry has entered a stage of transformation and upgrading.At present,the overall business model of the industry is still "extensive" operation,and enterprises cannot understand the characteristics of their user platforms and cannot use data to drive the transformation and upgrading of enterprises.By analyzing user behavior data and constructing a profile model,user profile technology can mine typical user characteristics.Therefore,building a corresponding user profile system can provide assistance to the development of enterprises in the industry.This thesis starts from the current predicament of the Internet access service industry,analyzes the service processes and business needs of the industry,determines the three dimensions of basic attributes,consumption attributes and interest attributes to describe user characteristics,and with the help of big data technology,three user profile models have been constructed,on this basis,a user profile system for the Internet access service industry is realized.The main work of this thesis is as follows:(1)In view of the problem that the growth rate of users in the Internet access service industry is decreasing year by year,this thesis puts forward an operation strategy for maintaining active users,and builds a prediction model of user churn.By referring to the concept of user return visit rate in website analysis,the period of user churn in the Internet service industry is defined,so the churn user is identified,and use feature engineering to process the data,Then,a method of user churn prediction based on the Stacking ensemble strategy is proposed.The experimental results show that,compared with the single classification algorithm,the accuracy of the churn prediction of this algorithm is improved,indicating that this method has better churn prediction performance.(2)Aiming at the problem of user marketing in the Internet access service industry,this thesis proposes to subdivide user value,adopt corresponding operation strategies according to different user values,and construct a user business values model based on the RFM model.Use clustering algorithms for data analysis to divide user groups.Moreover,the defect of the KMeans algorithm depending on the initialization is improved.After experimental analysis,the improved K-Means algorithm has greatly improved,which can improve the performance of the user business values model.(3)Aiming at the problem of user experience in the Internet access service industry,this thesis puts forward an operation strategy to provide personalized services based on user interest,and builds a user interest model.By analyzing the characteristics of the corresponding user interests,a hierarchical interest tag library is constructed,and on the basis of this,the users' online behavior data is modeled.In the keyword extraction of webpage content,an extraction method based on TF-IDF is proposed.Through experimental comparison,the improved method has better performance and can be better used for interest modeling.(4)This thesis implements the user profile system.The user profile system will collect and sort the required user profile data,analyze and process the data,tag the behavior data,and finally visualize the constructed user profile.Through system testing,the availability of the system is verified.