A Topical Study on the Web Spam

In this paper, we study the topical characteristic of spam hosts. To categorize spam hosts, we extract link spam structures from multiple time snapshots of Japanese Web archive using graph algorithms. Next, we define several spam topic categories and classify spam hosts in those structures into such spam topics using their uniform resource locator(URL)s and a machine learning approach. We analyze the spam topic distribution on the Web in different years and observe the change in spam topics through the time.