Analysis of Clustering Technique in Android Malware Detection

Abstract

Mobile computing is an important field in information technology, because of the wide use of mobile devises and mobile applications. Clustering gives good results with information retrieval (IR), It aims to automatically put similar applications in one cluster. In this paper, we evaluate clustering techniques in Android applications. We explain how we can apply clustering techniques in malware detection of Android applications. We also use machine learning techniques in auto detection of malware applications in the Android market. Our evaluation is given by clustering two categories of Android applications: business, and tools. We have extracted 18,174 Android's application files in our evaluation using clustering. We extract the features of the applications from applications' XML-files which contains permissions requested by applications. The results gives a positive indication of using unsupervised machine learning techniques in malware detection in mobile applications using a combination of the application information and xml Android Manifest files.