Design And Implementation Of Mobile Phone Virus Mining System Of Closed Sequential Pattern Based On Spark

With the development of the Internet and the popularity of smart phones,mobile phone viruses are increasing and the types are more abundant.Mobile phone viruses not only destroy user data but also steal user privacy,thus which causing great harm to users.So how to monitor and prevent mobile phone viruses effectively has become a widespread concern of society.As the viruses cause the abnormal behavior and hide certain regularity,the data mining technology can be used to find out the representative mobile phone virus behavior features to detect virus accurately and actively.Currently,applying data mining approaches to mobile phone virus mining has the following problems:(1)The stand-alone version of the mobile phone virus mining system is inefficient in processing large-scale data;(2)Most existing mobile phone virus mining methods do not consider the timing relationship between virus behaviors;(3)Most mobile phone virus mining systems have simple function.Users can not use the system autonomously nor analyze intuitively the effectiveness of mobile phone virus mining results and test results.To handle these mentioned problems,this thesis focuses on the closed sequential pattern mining technology based on Spark,which is a distributed computing platform.Besides it,this thesis designs and implements a mobile phone virus mining system based on the technology.The main contributions of this thesis are:(1)Analyze the core working principle of the Spark framework.Deploy the distributed Hadoop cluster and the Spark cluster on three computers with Ubuntu system;(2)Analyze the implementation principle of the stand-alone closed sequential pattern mining algorithm CloTSP.Design and implement the parallel closed sequential pattern mining algorithm S-CloTSP in Scala based on Spark platform combined with the characteristics of the CloTSP and big data processing.The effectiveness and parallelization efficency of the algorithm are demonstrated by experiments;(3)Apply S-CloTSP on mobile phone virus mining,designing and implementing the mobile phone virus mining system.On the one hand,the system applies closed sequential pattern mining on sequenced virus behavior,which not only considers the timing relationship between behaviors,but also effectively solves the problem that the frequent sequence sets are redundant.The background operations based on Spark make the system suitable for massive data mining by considering its mining efficiency.On the other hand,the system provides users with a visual web interface.Users can explore potential mobile phone virus behavior features by simply clicking and dragging operations,and detect unknown data based on feature sets.The system quickly and intuitively expresses the test results in a rich graphical form,which helps users to quickly find their own conclusions in the data;(4)Finally,the system is tested for its performance and function ability.The experiment shows that the system is effective and available,and is valuable in the field of mobile virus mining.