Research On The Correlation Analysis Between Andrews' Application Function Description And System Authority

The Android system provides a broad environment to developers and combines many excellent Google services. This makes Android one of the most popular intelligent terminal operating system. Meanwhile, some malicious applications appear in android market. They have certain security problems, such as stealing users’ privacy malicious deduction, etc. In order to ensure users’system security, Android uses permission-based security mechanism and demands that developers declaring permissions used in the application in advance. However, many applications will have excessive authorization requirements, thus brings security problems.Aiming at excessive authorization requirements of malicious software, we research the relationship between the function of android applications and permissions, and put forward relationship analysis method. This method obtain the model between applications’ function description of each category and permission by text analysis. The model can be used for future work such as researching rationality and security of permission declaration in android applications, and provides application market and android users with supports for application credibility assessment.In order to achieve the method, we explore some famous android app markets in and out of country, and determine the source of data retrieval based on characteristics of relative app data. After that, we develop a network crawler tool to fetch the data information from relative networks, including android Application name、Current Version、Required Android、Category and overview. On the other hand, we categorize the application and obtain permission description information from android official website. We do some pretreatment operation like word splitting and semantic extension on text information, and build the relationship model.In order to verify the validity of the proposed method and relationship model, we conduct confirmatory analysis aiming at a group of test application set. The result shows that the average recall, precision and F-measure of forecast result of relationship between application function description and android system permission are65.53%、33.85%and49.69%respectively. It means that proposed method and model can accurately reflect the relationship.