Evaluating spam detection techniques using ranking algorithm in email network at Hanoi University

Vu Minh Tuan, Tran Quang Anh, Ha Quang Minh


In this paper, four spam-filtering approaches based on user ranking in the mail networks: Clustering, Extended Clustering Coefficient, PageRank Algorithm and Weighted PageRank Algorithm are analyzed. We also propose a couple of fully worked-out datasets from the email network of Hanoi University against which the experimental comparisons with the respect to the accuracy of email user ranking and spam filtering are conducted. The results indicate that PageRank Algorithm and Extended Clustering Coefficient approaches are better than others. The rate of true detection is over 99.5%, while the failed alarm remains below 0.5%.


Spam detection, email network, clustering, PageRank algorithm, user ranking

Full Text:


Journal of Computer Science and Cybernetics ISSN: 1813-9663

Published by Vietnam Academy of Science and Technology