Search Engines Playing 20 Questions

I keep harping on it over and over (and OVER) again, but the truth is that search engines really do want to judge your site the same way that a human being would. It's fairly easy to point out examples of how this has worked in the past (specifically with the maturation of link analysis) but sometimes it's tough to make guesses about what direction things will go in the future. After all, there are eleventy billion ways that a human judges a site right now that a search engine can't. Well, Matt McGee has pointed out one possible option that could bring search engines more in line with the Pinocchio Effect in regards to local search.

Matt points to Bill Slawski's latest writing about a Google patent application that deals with category synonyms for some background before explaining a very interesting way that Google could use human generated data to create a sort of latent semantic indexing option for properties like Google's local search.

It seems that Google Base is playing a game of 20 Questions to try to understand more about the products being uploaded into its database.

Basically, what's happening is that Google Base is taking uploaded data and trying to expand on things like product and business descriptions. For example, if you upload something that says you offer "car rental" Google might ask you if "car rental is the same as car hire" because Google knows that "car hire" is the term used in the UK.

But Matt takes it a step beyond that and looks at how Google could use this technology to help businesses better explain their offerings in the Local Search database...

Matt says he's just thinking out loud, but I think this is a perfect example of how search engine algorithms and search engine databases could continue moving toward that goal of becoming a real boy. (No, search engines aren't girls...girls tell you what you need to know before you ask.)