Interesting little piece in New Scientist about a bunch of Cornell scientists who have developed some algorithms to track the frequency of words in documents in order to identify social trends. Lots of crossover with the stuff Cameron Marlow (the bloke behind blogdex) showed me in MIT a couple of years back, and with the work that Noll's brother does; plus a hint that Google's PageRank algorithms also use word frequency spikes for Google News.
Word frequency trends -- especially when used in conjunction with the clustering ideas Noll talks about (c.f. the idea of a search for "Challenger" coming up with different results grouped by coincidence of other words: boxing, tanks, space) -- must be a far better way of finding related content, and of establishing relationships, than absolute "snapshot" keywords, Autonomy, etc.
(From New Scientist)
Thursday, February 20, 2003
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment