Bill has an excellent, and very detailed, post on Google’s search query refinement patent.
“I find the most interesting thing about this approach is how user data, through log files and cached information, is incorporated into results.
There are probably some hints in the document for people creating content on web sites, on how to best show up in some of these alternative results. Taking a careful look at words you think people will search for, and what appears in Googleâ€™s results for those search phrases may tell you something about how your pages will be treated by this search refinement approach“.
I find that most interesting too, Bill. This reminds me of the tilde search, i.e. the placement of a ~ before a Google search query in order to see the related synonyms and webpages within the cluster. Knowing the search keyword terms is great, but there’s a lot of gold to be be found in knowing the keyword terms associated with that phrase, and, more importantly, the keyword terms Google associates with that phrase.