Commenting on Google’s Universal Search press release last Thursday, Gord Hotchkiss tied in Google’s personalization efforts to their ability to connect different silos of information:

The key to universal search results is an on-the-fly algorithm that looks across all of Google’s information sources and prioritizes and ranks all the items coming from these disparate sources based on the user intent. Now, it’s in those last five words, “based on the user intent” that the really important piece of this comes out. Just a few weeks ago, I interviewed Marissa Mayer about the inclusion of Web history in the dataset to calculate personalized search results. This just gives Sep Kamvar and his personalization algorithm a lot more to chew on as they determine user intent.

Gord’s a wise guy. He realizes that if you’re trying to prioritize images and video in addition to text results, and base their ranking on what you think the user is looking for, you’ll have to have a pretty powerful mechanism for ‘disambiguating user intent.’

What a beautiful phrase, ‘disambiguating user intent’. If you love language as I do, you have to appreciate it, which is why I’ve repeated it three times so far. In addition to its beauty, though, it underlies the core premise of Google’s personalization technology: in order to be effective, you have to figure out what the user is trying to find.

This may sound like an obvious mandate for a search technology. But what if you could take the clues provided by search history and demographics, and overlay a deep understanding of the users themselves? MyWebDNA’s validated results came back showing a 14% increase in relevance of search results based on a user’s core purpose and values. Those results indicate that understanding the users could prove to be a powerful means of disambiguating their intent (couldn’t resist).

In other words, you can predict as much or more about someone’s behavior from understanding who they are as you can from understanding what they want.

This is pretty interesting from an algorithm approach, because what you want can change moment to moment, while who you are tends to shift over far longer periods of time. The two combined can result in relevance on a much deeper level.

If Google’s serious about users seeing more accurate results, more often, don’t you think this is something they should explore?

This entry was posted on Monday, May 21st, 2007 at 10:28 am and is filed under Google, Relevance, Search. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.