Design And Implementation Of Mobile User Behavior Analysis Mechanism

With the revolution of mobile devices and telecommunication technologies,traditional network cannot meet the demand of accessing Internet at anytime and from anywhere,which gives birth to the mobile Internet.The mobile Internet brings in the business models,which provides the opportunity and the challenge for every walk of life.Under the mobile Internet environment,enterprises may focus more on user behavior s,which can guide making business decisions.We propose in this thesis an analysis mechanism considering mobile user behavior,which takes into account user moving behavior and interest behavior at the same time,and the specific works are as follows:Firstly,we come up with a time window based trajectory mining algorithm for mining user moving trajectories from user moving log data;Secondly,we design a modified trajectory clustering analysis algorithm for individual user moving behavior construction,with the definition of cluster closeness;Thirdly,we put forward a large-scale trajectory data clustering algorithm,more suitable for group user moving behavior analysis;Fourthly,for user interest behavior analysis,this thesis utilizes Naive Bayes algorithm to analyze individual mobile user interest behavior,and design a modified Naive Bayes classifier to solve unevenly distributed webpages for different themes.Finally,based on mapreduce programming model,we design a distributed Naive Bayes classifier,which can be used to process large-scale webpage data.The proposed mobile user behavior analysis mechanism is implemented and validated in Hadoop platform.The result shows that,the proposed mechanism can mine mobile user moving behavior and interest behavior effectively.Moreover,compared with classic DBSCAN algorithm and classic Naive Bayes algorithm respectively,the proposed mechanism in the thesis can analyze user moving behavior and interest behavior more accurately.