Towards Efficient Splog Detection via Local Web Graph Analysis

The aim of the current project is to research ways of dynamically “crawling” the Internet to collect enough information to be able to make a confident conclusion of whether a particular wep-page or blog is spam. This is proposed to be done by continually maintaining a `backbone' subset of the set of all web pages containing the pinging blogs and trusted websites, and doing a minor crawl based on the web page that is to be analyzed. Obtaining the representative set of trusted nodes and developing the crawl that will result in a reasonable neighborhood of the set in question are the main goals of the project.

Evgeny Skvortsov
Faculty Supervisor: 
Dr. Andrei Bulatov
British Columbia