Abstract: As is known to all, vulnerabilities play an extremely important role in network security now. Accurately quantizing the exploitability of a vulnerability is critical to the attack-graph based analysis of network information system security. Currently the most widely used assessment system for vulnerability exploitability is the common vulnerability scoring system (CVSS). Firstly, the exploitability scores of 54331 vulnerabilities are computed by using CVSS. Then, statistical analysis is performed on the computed exploitability scores, which indicates that CVSS lacks diversity, and more diverse results can help end-users prioritize vulnerabilities and fix those that pose the greatest risks at first. Statistical results show that the scores are too centralized as well. Finally, taking into account the disadvantages of CVSS, we study the influence factors of vulnerability exploitability, and demonstrate that the types of a vulnerability can influence its exploitability. Therefore, we consider vulnerability types as one of the influence factors of vulnerability exploitability, and use analytic hierarchy process to quantify it, and propose a more comprehensive quantitative evaluation system named exploitability of vulnerability scoring systems (EOVSS) based on CVSS. Experiments show that the diversity of scores computed by EOVSS is four times that computed by CVSS, and EOVSS can more accurately and effectively quantify the exploitability of a vulnerability in comparison with CVSS.

Key words: vulnerability, exploitability, vulnerability type, analytic hierarchy process, quantification