期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2018
卷号：96
期号：20
出版社：Journal of Theoretical and Applied
摘要：The great shipment of Android mobile devices throughout the world has surged the application develop-ment. Indirectly, this scenario had invited the malware creator to be in-line with the technology evolution. One of the threats is the leakage of privacy data and it is a serious subject. To overcome this, the Android application usually being examine through static or dynamic analysis. In static analysis approach, re-searcher commonly considered combination static features to identify the benign and malicious applica-tion. This paper presents a proof of concept on classifying Android benign and malicious apps by its appli-cation category. At the same time, this paper proposes a new framework for malicious detection focusing on the leakage of user privacy using minimum number of the request permissions and API calls features. Several machine learning classifiers with several training and testing percentage applied in this study to compare the accuracy. The result show that, applications in same category reported more accurate per-formance in identify malicious apps compared to non-category based. By applying features ranking and information gain features selection, Random forest classifier with 10 folds cross validation for both �Book and Reference� and �Personalization� category achieved higher true positive rate also lower false positive rate.
