Spam Detection in Link Shortening Web Services Through Social Network Data Analysis

Abstract

Twitter is one of the most popular social networking and micro-blogging Web sites in the world. TinyURL is a URL or link shortening web service used by Twitter. Recently, it is being exploited by spammers as a podium to transmit malicious information. TinyURL spam detection in Twitter is a challenging task. In this paper, an efficient scheme is proposed to detect spam in the TinyURL with particular focus on Twitter tweets. A set of features are first extracted from the tweets. This feature set is analyzed to select a reduced set of features. The reduced feature set is fed as input to train three classifiers, namely simple logistic regression, decision tree, and SVM. The classification results show that the SVM classifier has the highest accuracy in detecting spam in TinyURLs.