Analysis and Classification of Danger Level in Android Applications Using Naive Bayes Algorithm

This paper considers danger level classification of Android applications based on permissions and vulnerabilities by using Naive Bayes (NB) algorithm in order to assist and inform users whether an application is safe to use or not. With the increasing development and use of Android, unfortunately, malicious software (malware) and malicious applications are also beginning to increase. Many methods have been proposed to protect Android, however, they are only able to detect or classify Android applications against malware based on permission. This kind of approach is still considered less effective, because there is no information in classifying the danger level of an Android application, be it malware or goodware. To overcome the problem, this research classifies the danger level into three categories namely, safe, suspicious, and dangerous. The accuracy obtained from this research is 97.2%. To our knowledge, this is the first and only work to use danger level classification of Android applications based on permissions and vulnerabilities.