Massive Website Access Behavior Analysis Based On Pattern Minning

With the rapid development of 4G mobile Internet,mobile Internet content services are further enriched.The traffic data generated by users using mobile Internet also contain more information,including user behavior trajectory,living habits and so on.How to analyze the meaningful information from the massive data of mobile Internet has become a hot research direction.This paper mainly uses pattern mining to analyze users' website access behavior.Firstly,frequent pattern mining combined with appropriate single segmentation strategy is used to explore the relationship between websites in the access process.Then aiming at the lack of directional characteristics in the associated patterns,the sequential pattern mining is innovatively introduced to obtain frequent website sequences,and then to obtain the directed correlation between websites.Finally,the order funnel analysis of the website sequence is carried out,and the conversion rate of each website is used to measure the strength of the directed correlation.The main contents of this paper are as follows:first,fr-equent pattern mining combined with different ticket segmentation strategies is used to explore the relationship between websites,and the experimental idea is applied to the malicious websites binding detection in the process of website access.Secondly,the sequential pattern mining method is introduced and we modify the sequential pattern mining algorithm to analyze the website access behavior.By using the method of comparative experiment,we obtain the ticket segmentation strategy suitable for the analysis of website access sequence,and apply the segmentation strategy to deal with mobile Internet traffic tickets,and obtain the sequence patterns of website access behavior.This is the result of directed association rules between websites.Thirdly,we choose an ordered funnel analysis method which is suitable for sequence analysis,and optimize the original ordered funnel analysis algorithm so that it still has good performance in analyzing massive data.According to the characteristics of website access,the time window in the definition of ordered funnel analysis is modified to the interval window,and the conversion rate of each website in the sequence patterns is obtained,that is,the degree of directed correlation between websites.