Next Gen News

like Nostradamus

Nautilus supercomputer is brilliant in the prediction of events, provided they have passed

The best predictions are always a posteriori. Nostradamus became famous for writing a bunch of crap wave of the future, and with the amazing human brain's ability to detect patterns and links - even where none exists - to do the rest. His legacy is a large number of prophecies that are absolutely brilliant in the prediction of events, provided they have passed. Nostradamus now a competitor of silicon, but on seeing French generates around 900 or more of his quatrains, the University of Tennessee "Nautilus" supercomputer is capable of spitting out millions and millions of predictions - that sufficient to maintain an army of hard buckets from here to eternity. Nautilus rolled through hundreds of millions of news articles, the application of algorithms for sentiment analysis and detection of a place name to take a theme and are correlated with a place and a state of mood. The theory is that if "Brad Pitt"

is mentioned in many articles in French with positive words like "big"

"brilliant"

then the French people love it, but if

"Mubarak"

is associated with

• "bad"
• "horrible"
• then it's time to get up and pitch fork Egypt the manufacturers.
• The researcher behind this, Kalev Leetaru, the
• University of Illinois, Institute of Computer in the Humanities, Arts and Social Sciences speculated in a recent document that mining of this information, the system could predict anything from

"forecast imminent conflict to provide views on the location of fugitives."

These statements were amplified by the BBC, they score a lot of bonus points for a link to the actual paper (!), But immediately lost the use of "scientists say" is refer to a single investigator (probably forgot to update your model) and explain that it is not likely to have a better chance of predicting the next evacuation, Hosni Mubarak, this supercomputer is to anticipate future revolutions. I'm not a big fan of sentiment analysis, because every time I try to use it in a concrete situation the results are repeated, of course, spectacularly wrong. To illustrate the problem, here's one I just made reference to "Dorries" - a term that some of you may know -. In Twitter (you can see a longer version here)

algorithm sentiment analysis estimated that about half of the mentions are positive. The algorithm sentiment analysis is to talk nonsense. Tweets algorithm that rates as positive the fragments "Dorries" include the following views:

"I think we should abolish Dorries"

"Dorries lose his seat in the contouring"

"Keep your hands dirty my dog ??[****] horrible"

Find best price for : --Nostradamus--