关键词: Android系统, 多特征, 朴素贝叶斯, 恶意软件

Abstract: Traditional malware detecting methods on Android system based on permissions have a high detecting rate, but also have a high false positive rate. The characters of the methods based on API Calls are hard to extract, so the method is difficult to apply to mobile platform. This article proposes an Android detecting method based on the Naive Bayes algorithm with the characters combination of permissions and resources. The characters are easy to extract and the experiment shows that the method has a low false positive rate. It collects 4, 396 malwares and 4, 500 benign applications and randomly selects five groups of malwares and five groups of benign applications as the test sets. The contrast experiment on the test set with permissions and multiple characters is done. The result presents, compared with the method based on permissions, the method based on multiple characters, can remarkably reduce the false positive rate. The method based on multiply characters is better.

Key words: Android system, multiple characters, Naive Bayes, malware