K-means and Principal Components Analysis Approach For Clustering Malicious URLs

Security flaws in the world of cyber security are a major concern with the advancement of internet technologies. Malicious URLs are increasing with the number of web links in the world. Developing techniques to classify blacklisted malicious URLs is one of the most focused solutions in the security community. In this paper, we developed a clustering approach to find the similarity into different malicious URLs.The effectiveness of our technique was evaluated using the ISCX-URL-2016 data set, which contains over 110,000 URLs. The collection of necessary features group data is included in the ISCX-URL-2016 data set. Therefore, the data were prepossessed and scaled, then two most commonly used lightweight algorithms are implemented: k-means and principal component analysis. Experimental results exhibit by combining the two algorithms has produced several evaluation results of identifying new potential attack categorizes.