Suspicious URL detection system using SGD Algorithm for twitter stream

Twitter is a one of the most popular social networking site used by millions of people in the world. As the usage growing rapidly in the recent years, attackers are concentrating more on twitter to gather personnel data and made changes to. It leads to the diminishing the privacy of the users. The attacker tweets suspicious URLs on the user’s timeline. These URLs contains spam, phishing and malware distribution. There are many traditional URL detection algorithms which take more time and resources. These algorithms use html content, dynamic data, and lexical features to detect the suspicious URL. We propose a new detection algorithm which uses correlation of URL redirection to detect suspicious URLs. The redirection the URLs repeatedly send to same link again. We find frequent repeated URLs as suspicious. We use an online training algorithm called stochastic gradient descent algorithm to train the system. It can be used a classifier to detect the benign or malicious URL. We are also adding another module called alert system which will send all the suspicious URLs to the twitter server. The proposed method improves the performance of the detection system.