Personalized Search – User Manipulated Search Results

by TheMadHat on March 7, 2007

Welcome to Part I of a long awaited collaboration between TheMadHat and TheGypsy. We’ve been digging deep into personalized search lately, and there has been a lot of chatter about the subject around the SEO world. In part I of our collaboration, I’m going to go through the latest patent on the removal and manipulation of personalized search, and what this means for you, the SEO mastermind. You can see the full patent here.

This patent is a fairly simple one (when it comes to the world of patents anyway). Essentially it goes through the different methods a user could remove unwanted results, and how this data could be used in ranking results. Let’s break it down:

Each search result would have an option to immediately remove the result for that search if the user deems it spam, irrelevant, or offensive. By default, the result would be removed only for that search session. The next time you jump back on and perform the same search your results will be standard (assuming other personalization factors and not influencing the listings). This is the basic, bare bones example and would cause the least problems from and SEO standpoint.

The second option is one that could cause a little more turbulence. The user would have the option to remove that specific result from all future searches. Keep in mind we’re talking about only one listing at this point. Bye bye. You’re now invisible to this user for this particular page. It will never show up again for that search.

The third option is the killer. This option allows you to remove this result, and all documents associated with this result. Bye bye big time. Your entire site is now invisible to this user from now on.

What does all this mean to you? Not a big deal you’re thinking? Well think again. Sure, users removing you from their results were not likely to buy anyway. Not a big loss there. However, let’s look at these specific sections of the patent:

“…aggregating information regarding documents that have been removed by a group of users; and assign scores to a set of documents based on the aggregated information.”

“…determining a remove list score associated with the documents in the set of documents based on the aggregated information;”

Now we’re talking. Each result will be assigned a “remove list score” and rankings will be determined by “link-based score, the information retrieval score, and the remove list score”. My first thought was that this would be easily gamed by all you spammers. Automated account creation, mass removals, etc. Obviously Google has thought of that as well:

“identifying a set of legitimate users and a set of illegitimate users; and collecting information regarding documents that have been removed by the set of legitimate users.”

Now it doesn’t go into detail about how they define a legitimate user. My theory (guess) is that they will look at: length of time on the account, number of searches performed, number of removals done, etc. Sort of a trust factor for the account in question.

What does this mean to us? The ease of removing a site will be a big factor, but if it’s easy I can see lots of users removing stuff in droves. If your site sucks, people will remove you and this will cause your rankings to plummet. It just emphasizes the fact that you need to have content that people will eat up and demonstrate you are an expert in your niche. A professional, well designed site that makes your users comfortable is also going to be a factor. Obviously Google doesn’t know what your site looks like, but it can make some assumptions based on removal data. Goodbye MFA’s and a ton of affiliate garbage.

Make sure you’re ahead of the curve with unique and compelling content and you should reap the rewards. Seems like a standard sentiment now days.

Similar Posts:

Comments on this entry are closed.

Previous post:

Next post: