| The rapid development of mobile applications has promoted the development of mobile Internet.The massive mobile applications make the network connection in the application have security problems that can not be ignored.At present,the introduction of too many third-party libraries in application development leads to the introduction of many unknown network connections,which eventually leads to the occurrence of malicious events,which not only affects the user experience of Android applications,but also divulges user information,and even eventually causes the loss of property.There are many and messy places where network connections are generated in the application,and all network connections mixed together are considered to be generated by the application.Therefore,this paper proposes to divide the responsible person of each network connection in Android application,and divide the network connection more finely,which can prevent or locate the specific malicious network connection as soon as possible.Finally,the results are applied to the field of third-party recommendation,and good recommendation results are obtained.The main work achievements of this paper are divided into the following four aspects:(1)The method of extracting network connection in native layer is designed and implemented.The network connection of native layer has been ignored all the time.This paper studies the design and implementation of the network connection extraction system of native layer,and reveals the relationship between the network connection of native layer and the size of application,the type of network connection of native layer and its security status.(2)A method of fine-grained classification of third-party network connections in applications is designed and implemented,which is accurate to the library.After using the above network connection extraction system to extract the network connection information of nearly 500000 Android applications,first divide the network connection into third-party module network connection and application own module network connection,and then analyze the network connection resource library of each third-party library according to the characteristics that the third-party library has been called for many times.In this paper,a total of 1259 third-party library network connection resource libraries are generated.Combined with the network connection security detection results,it is found that there are unsafe network connections in the network connection resource libraries of 27 third-party libraries.(3)This paper detects the security of network connection of 500000 applications,analyzes the security of network connection at all levels of applications combined with the results of fine-grained classification,and reveals the current situation of network connection security of Android applications.(4)Based on the analysis of the network connection security results and functional similarity of the third-party library,this paper recommends the third-party library for 12 categories of Android applications,and achieves high accuracy. |