remember a while ago when i was telling you about these pigeons that were actually blogging? did you also know that pigeons think in logarithms? well, this is another bird-themed blogging bit, but a much more sophisticated take on how information gets organized…and as usual, this is something that the search engine troika could learn from (along with technorati and the other blog engines)

the problem: my blind dog keeps shitting in my neighbor’s yard…no, i’m kidding, lemme start over: the problem: too many blogs and news sources out there, too many rss feeds to manage and track, too many stupid and primitive ways to rank and sort the blogs (tagging, credibility, reliability, etc)…if you’re like “what’s rss? does it look like a mole but hurts?” then please call customer support…

…in a bit by the new scientist (best mag ev-ah) called “a little bird told me” (got it in print, can’t find the online link), there’s a summary describing the work of xiaohui cui and his chums at the oak ridge national lab in tennessee. they’ve got this system full of ‘virtual birds‘ and per the article, “each bird carries a document, which is assigned a string of numbers…docs with a lot of similar words have numbers strings of the same length and a virtual bird will only fly with others of it’s own ’species’ (same string lengths)…so when a new article appears, software scans it for words similar to those in existing articles and it files it into a flock (or creates a new one)”

…and yeah, it’s already working, and the next big thing is allowing users to click on birds to see its content (document)…hypothetically, this could allow for a user to discover associations through a machine process that would never have come to our normal brains without the flocky algorithm, all of which could beat the absolute hell out of services like all of the news alerts out there and turn us all into superior analysts, liberating us from some of the underlying research tedium…the whole group behind this is part of the ornl’s sw group

you can view a demo of cui’s flock based clustering algorithm by visiting his research summary site, but i can’t find the interactive demo for rss processing…in fact, call me paranoid but it seems to have vanished or something - and then consider that it’s under the auspices of the same bright mind who created vipar for osint application, and you have to wonder if maybe this kind of information processing power is more useful for government application than consumer…hmmmm...am i just being paranoid? why else would ornl make it so difficult to find the demo that new scientist got to view when writing up the summary? (there’s a whole swarm algorithm conference coming up, maybe this kinda stuff will be on display there - anybody going?)

Powered by Gregarious (21)

Share This

Some similar nonsense, if you like that kind of thing:

Links that were in this post, but only if there were any: