Detection of phishing webpages based on visual similarity

Reviewer: Shannon Jacobs

Phishing is a label for certain strategies used by criminals to steal personal information via the Internet, and it is a rapidly growing problem. The classic approach involves using spam email containing "bait," typically some story about why the user should visit a Web site and "revalidate" their personal account information?while actually revealing that information in a counterfeit Web page created by the phisher. This report briefly summarizes three useful metrics the authors have evaluated for the detection of such counterfeit Web pages. The unstated premise is that this applies to phishers who are targeting "experienced" victims who have some familiarity with the authentic Web sites, so the visual similarity is critical to their phishing. (This premise is actually questionable, since experienced users are more likely to spot phishing attempts regardless of what the Web sites look like.) The main drawback of this report is that it doesn't present sufficiently detailed information about the results, but the limited space excuses that. A less serious (but less excusable) problem is that the English is significantly flawed, though most of the meaning can be teased out with a bit of interpretive effort. Online Computing Reviews Service