This poster paper was published in the recently concluded WWW 2008 conference in Beijing.The author suggests that spam blogs (splogs) form co-citation clusters because they share advertisement links between each other.The experiments were based on 691,674 blogs, collected during a week in 2007, from the Japanese blogosphere. The author reports that high out-degree blogs are usually spam blogs(95% of the time). An iterative spam traversal algorithm is implemented to extract spams using co-citation cluster analysis of known spam blogs(seed set).The seed set was automatically generated using high out-degree pages and pages containing adult and commercial keywords.
Comments :
* He does not mention what is meant by high out-degree.
* Method seems simple and intuitive, but more exhaustive experiments would provide more insight.
. For the mathematically inclined, I achieved a ten-fold cross validation accuracy of 80.48%. For details, read the complete report.
), I was kinda skeptical about the prospect of our product. Working weekends, overtime, we slowly trudged towards our goal. Keeping the emotional rants aside, I just want this wonderful feeling to sink in. Obviously this is not a finished product and we are continually trying to improve it. We are very well aware of the stiff competition that we face, neither do we have the resources nor the financial clout to compete directly with the other G 
,but that does not bother us. As Randy Pausch so rightly said, “Do the right thing and the good things will come to you…”, Thats all we are trying to do, “Keeping it real and Keeping it local”. In some future posts I’d discuss some of the cool features of our product and may be some secrets about ranking higher in Genieknows search 
