View Single Post
Old 18th Oct 2004, 11:25   #14
wshaw
Senior Palimpsester
could do better
 
wshaw's Avatar
 
Join Date: 13 May 2003
Location: Brighton, uk
Posts: 1,748
Default

Quote:
Originally Posted by John Self
It got me right, but the science seems distinctly dodgy. It does it all based on a few dozen 'keywords' each of which it defines as either male or female. So "the" and "a" are male, but "was" and "and" are female. According to the statistics, out of a quarter of a million samples, it has got it right 62% of the time and wrong 38% of the time - not very impressive really since it has a 50/50 chance if it just plucked a sex at random.
Got me right too, for non-fiction and fiction. Which I was a little depressed about.

I would have thought that the 62/38% split is, as they say, statistically significant though...
wshaw is offline   Reply With Quote