Design And Implementation Of Data Mining System Based On User Behaviour

Data mining is a very popular topic now,and has also been widely used in actual industrial production,and has achieved very good achievements.Because of the strong development of web technology,mobile applications have penetrated into the daily life of the people.They have brought a lot of convenience,and also brought huge user groups and operating profits to enterprises.Accordingly,users use the services provided by the enterprise application system,and the generated user actions and operation log data are also gradually increasing.By analyzing and mining the information hidden in the log data,some interesting patterns can be obtained,and these interesting patterns are of great significance for analyzing the requirements of users and evaluating the effects of products.At the same time,a good system service can better improve the system users retention rate,explore potential user groups,and bring huge profits for the company.In this paper,through the Hadoop large data technology,combined with the traditional web mining technology,the system generates logs for statistical analysis.This paper describes the present situation of the system in the domestic and foreign technical research and development of mining user behavior log based data,and analyzes the core business process requirements,and analyzes the system in function and performance requirements.The system mainly consists of three subsystems,such as job management,data analysis and data display,etc.at the same time,the functional requirements analysis,design,implementation and testing of the three subsystems are carried out.Because of the primitive user behavior,the operation record is rough,so it needs to be cleaned,extracted,transformed and loaded.At the same time according to the traditional data mining algorithm based on Hadoop big data development platform for parallel research and optimization,we can devide people into serveral cluster through user behavior log analysis and user interest measurement by fuzzy clustering algorithm,while the use of sequential pattern mining algorithm,that is apriorall,GSP,Prefix Span algorithm to analyze the behavior of the users access pattern,theme and content,and analyze user access preferences and interest model,and according to the model of personalized recommendation to provide good service for users,users can quickly locate their interested content.Web system display module uses LNMP technology to display data and analyze statistical results through various charts and forms.