Towards Measuring Apps' Privacy-Friendliness

Abstract: Today's phone could be described as a charismatic tool that has the ability to keep human beings captivated for a considerable amount of their precious time. Users remain in the illusory wonderland with free services, while their data becomes the subject to monetizing by a genie called big data. In other words, users pay with their personal data but the price is in a way invisible. Poor means to observe and to assess the consequences of data disclosure causes hindrance for the user to be aware of and to take preventive measures.Mobile operating systems use permission-based access control mechanism to guard system resources and sensors. Depending on the type, apps require explicit consent from the user in order to avail access to those permissions. Nonetheless, it does not put any constraint on access frequency. Granted privileges allow apps to access to users' personal information for indefinite period of time until being revoked explicitly. Available control tools lack monitoring facility which undermines the performance of access control model. It has the ability to create privacy risks and nontransparent handling of personal information for the data subject.This thesis argues that app behavior analysis yields information which has the potential to increase transparency, to enhance privacy protection, to raise awareness regarding consequences of data disclosure, and to assist the user in informed decision making while selecting apps or services. It introduces models and methods, and demonstrates the risks with experiment results. It also takes the risks into account and makes an effort to determine apps' privacy-friendliness based on empirical data from app-behavior analysis.