slawski - 10:53 pm on Jun 17, 2010 (gmt 0)
It was a surprise getting a call from Kenneth Li of the Financial Times, but we spent about 40 minutes or so on the phone talking about the Google patents. I asked some questions as well and answering some.
He found out about my post from another journalist who tweeted it, and after reading it brought it to his editor. It seems like this is a topic of particular concern to journalists.
Both he and his editor were wondering if the patent(s) meant that Google might get into the content generation business themselves, or provide a way for web publishers to become a little more competitive against some of the digital media sites like Demand Media that use algorithmic approaches to identify content for informational adsense-type pages.
There are two patents involved, an earlier one that has been granted, and a continuation that adds a little to the first and is still pending.
The patents cover looking at queries that have some baseline level of popularity and the quality and relevance scores of the pages that show up in results for those queries.
When Google identifies a query that has unserved or underserved results, it might look at related terms to see if there's a whole topic that doesn't have adequate results.
The new patent adds a little, including looking at queries where there might be adequate content in one language, but not in others.
The patents present a few different ways that they might inform web publishers of inadequate content, including telling searchers at the time of a search, creating a query and topic search engine, as well as providing the information to advertisers.
As synchronicity would have it, Demand Media had a patent application published today titled Method and System for Ranking of Keywords for Profitability (US Patent Application 20100153391) which might provide a window into the kinds of things they look for when attempting to identify keywords to write about.
Unsurprisingly, those include things like search volume, how much it might cost to bid on the terms, how competitive the terms are.
But, there are some issues with some of the examples they raise on things like how competitive a site might be. For instance, they state in the patent description that they might look at sites that rank highly for a query, and check the Alexa rank for those sites. I don't think that will help much.
There could be some potential benefits for ecommerce as well as content creation/advertising from a system like the one described in Google's patents. If you ran an ecommerce site, and looked up "leather widgets" in an "inadequate topics and queries" search engine, and found that there were a lot of searches for "green leather widget" without any pages that offered them. If you sold blue leather widgets and red leather widgets, would you consider adding green leather widgets to your site?