查看文章

The security vulnerabilities in software systems can be categorized by either the cause or severity. Several software vulnerabilities datasets for major operating systems and web servers are examined. The goal is to identify the attributes of each category that can potentially be exploited for enhancing security. Linking a vulnerability type to a severity level can help us prioritize testing to develop more effective testing plans. Instead of using an ad hoc security testing approach, testing can be directed to vulnerabilities with higher risk. Modeling vulnerabilities by category can be used to improve the post-release maintenance and patching processes by providing estimation for the number of vulnerabilities of individual types and their severity levels. We also show that it is possible to apply vulnerability discovery models to individual categories which can project the types of vulnerabilities to be expected in near future.