because android phones are widespread and steadily gaining popularity. Unfortunately, such phenomenon draws attention to malicious application developers so that malicious applications are increasing rapidly. To detect these malicious applications, several researches have been done using machine learning. In machine learning methods, there are two ways which are dynamic detection and static detection to detect these malicious applications. The dynamic detection has a high detection rate but uses a lot of resources and takes long time to detect malicious behaviors. In addition, it is possible to infect malicious code after implementing detection method. On the other hand, the static detection has a small amount of resources and quick detection time but has low detection rate. To these weaknesses, we propose a malicious application detection framework on android market and an automatic extraction tool of malicious features for static detection. This framework is able to perform both detection methods using machine learning and is expected to perform a thorough analysis than previous framework. Also, the proposed automatic extraction tool is expected to help a code analysis for static detection.
Download Full PDF Version (Non-Commercial Use)