Loogle V2

Because I'm not much of a computer geek, I'd best describe what this is about by example:

Let's say Burt is an expert in underwater volcanoes, and currently his computer file organization has accumulated a high degree of entropy. Stored on one drive are his geologic surveys, in another far-off location are his biological journals, and his environmental conditions of volcanoes are in a third. Burt writes a paper touching and including all of these subjects, which he'll file under his finished papers in some other random spot.

Burt's computer search system, during idle time, parses the words in the paper, and associates significant words automatically in a relational database. It then communicates these significant words with Google or a probabilistic search engine like it. Google returns its results for likely hits of other significant words that are frequently associated with the originals back to the operating system. The operating system then goes and searches all local files for the Google suggested return hits, (and optionally saves a copy of that search session, which it passes on to a selected peer group, in Burt's case, trusted fellow oceanographers). All local files that turn in a match are linked or shortcutted or whatever in one nice, neat, delivered package, which Burt can adjust and trim to teach the neural system. Effectively, Burt's computer now understands that underwater volcanoes and tubeworms have something to do with eachother, from a topical standpoint.