Evaluating mobile malware by extracting user experience-based features

Abstract

The perception the user has about the performances of an application could determine the success of the application in the marketplace. The research community has made many efforts to understand out how to measure the user experience of mobile applications. In this paper we apply a set of features, typically used to evaluate mobile user experience (UX), with the aim at understanding whether there are differences in terms of usability between trusted and malware mobile samples. As a side effect we evaluate whether the feature set is useful to discriminate mobile malware. The experiment suggests that malware samples exhibit a better UX than legitimate ones. Furthermore we obtain, training several classifiers from UX-based features, a precision of 0.97 in malware identification.