Abstract: Web vulnerabilities develop in a dramatic speed ac-cording to the booming of IT technology and have become a serious threat to web security. Meanwhile, most dominant scanning approaches in current research community and commercial systems are low efficiency. In this study we propose a novel approach to improve the efficiency of web vulnerabilities scanner by designing the similarity comparison and template extraction. Therefore the novel approach can obtain the web structure dynamically and avoid capturing the same pages repeatedly. Experiments in this study show that our method achieves efficient detection of Web page vulnerability, meanwhile, it also has a high adaptability and flexibility.