Feb
14
Google AdWords Conspiracy and Taxonomies
Filed Under General, Google | Leave a Comment
In November of 2003 what was dubbed the Google Florida Update hit, with a massive uproar ensuing as websites got dropped into oblivion in the index en masse. Speculations immediately abounded that it was an “AdWords filter” or “money words filter” since it was right in the midst of the Christmas Holiday shopping season, while others called it an over-optimization filter and still others decided that the Hilltop Algo was being used, or that the re-ranking algo that was patented by Google had been implemented. With the holiday season upon us this year, it’s not unlikely that there will again be the AdWords blame game that’s played this year when the next “happening” takes place, which has to be imminent because Google’s been too quiet for too long with nothing changing. Like the calm before the storm.
Yeah, right! It was a company conspiracy to pick all the keyword words and phrases they knew were money-makers and apply a filter to all the sites pursing those to increase AdWords revenue during the shopping season, particarly in view of the upcoming IPO.
It would have been dangerous to let that go out to staff by company email in case in future some irate former employee who didn’t like the cooking went postal and exposed the whole scheme, so what they must have done is all dress up in their tin foil hats and have secret meetings in the back room of some local Mt. View saloon - just in case the Plex had been bugged by industrial saboteurs who had worked their way into the company to spy. Of course someone could have been wired, but ya gotta draw the line someplace, dontcha?
Meantime, it was stealthily added to their webmaster help section that they’d started to use stemming, so it was obvious they were doing something with words . Also, here and there we see evidence of changes or tests regarding clustering, where while it’s always been that when there are two pages from a site that are relevant one page would appear indented. Occasionally, we’ve seen pages appearing separately, without being indented for a brief season, with it then being reverted back. So that indicates something going on with sites and pages.
http://www.webmasterworld.com/forum3/21115.htm
It always gets me thinking that beyond keywords on pages, not only regarding Google, but other search engines as well. The engines have to see something related to semantic relationships within an entire site - and not only the site itself, but the sites and individual pages they link to and the ones that link to them.
Just a few of the references I’ve accumulated on LSI:
using lsi to classify search engine spam
http://www.webmasterworld.com/forum3/24383.htm
Using Semantic Analysis to Classify Search Engine Spam ( PDF Document)
HTML Version from the Google Cache
LSI Paper at Microsoft Research
Makes mention of using classification and taxonomies in web search
Web Search Using Automatic Classification
Paper at Stanford