Detecting app clones in Android markets using graph mining approaches
Muhammad Nurdin Bin Affandi
Date of Issue2017-05-17
School of Electrical and Electronic Engineering
With an estimated of 2.1 million people having access to smartphones, the number and variety of mobile applications (mobile apps) are increasing rapidly. These mobile apps provide much functionality and convenience to users and thus extending the capabilities of smartphones since its introduction. However, along with these benefits and advantages of mobile apps, it also allows new threats to surface that may pose a threat to the privacy and security of the users. For example, attackers may duplicate codes from legitimate Android apps and reassemble it with malicious codes that may do harm to users or introduce “purpose-added” functionalities that benefit these attackers. This project is, therefore, trying to address the issue of clone apps by detecting it through the method of graph mining. This report highlights the implementation of clone apps detection based on the approach of geometric characteristics of mobile apps called centroid using Python programming language to measure the similarity of methods between apps and draw a conclusion on whether an app is a clone or not. The app clone detection system implemented in this paper is tested on 260 apps collected and the 1,653,985 methods in it. The report will talk about the accuracy of the clone detection system implemented as well as the analysis of third-party library with respect to the app clone detection system.
Final Year Project (FYP)
Nanyang Technological University