Hot on the heels of the eagerly scrutinised Google Patent, an application filed by Yahoo has followed only 5 days later. In fact, this patent has been available since the end of March, but hasn't been picked up by the public until now - in many ways this is a measure as much as anything of how much more buzz Google seems to generate than it's rivals. The patent itself was filed over 3 months before Google's, back in what now seems a dim and distant 2003!
Those hoping for pistols at dawn as the two giants of search duke it out over patent infringement might be disappointed - by and large both cover quite different aspects of the search problem. Whilst the Google patent concerned itself mostly over pages searched, what they contained, how they change over time and how often people search for them, the Yahoo patent concerns itself much more with understanding users actual queries.
The document discusses "concept networks", "conceptual units" and "superunits", trying and to find and categorise the underlying ideas behind search queries, by looking at the language itself and the choices users make. In many ways, it is very reminiscent of Googles "synonyms" and "sets", but derived from user searches.
Although initially the language of the patent is very dense at first, a number of observers have pointed out how Yahoo have gone to great lengths to make the document easy to follow with examples and have speculated how the patent may be more of an exercise in PR than IP protection. Given the timing, you have to wonder!
But then again if PR were the goal, one of the most striking feature between the Yahoo and Google patents is - bizarrely - the quality of the diagrams. Although Google's are neat and well presented, Yahoo's look suspiciously like someone had a great idea and scribbled it down on a napkin. And then submitted the napkin. Ten out of ten for a groundbreaking algorithm, boys, but minus one gold star for neatness!
Another interesting feature of Yahoo's patent is that rather than concentrate just on the technological aspect, it makes explicit references to human editorial control to group keywords on a cultural level (like identifying groups of names as famous actors or singer). The human editor nature of Yahoo has long been its USP, and it's nice to see that it looks like that aspect of it's nature is very much intended to be preserved.
