In “traditional” IR systems, it is common to use a process in which search terms become better stated through iteration (based on analysis and improvement upon each search, until a satisfactory result is found). This is called Relevance Feedback.

The example here Google Chav, briefly illustrates the concept of Relevance Feedforward, which uses knowledge about the stereotypical likes of a user in order to tailor the results prior to the initial search, shortening the search cycle, and providing results that are more attuned to the searcher’s idioms.

One day, all search engines may have the capability. Writing about it here means nobody can patent the idea, ’cause this is now prior art.