Saturday, April 19, 2008

leximancer duel - izzard vs hicks

Barry Saunders' stuff has inspired me to have a look at some text mining tools. So I got a trial copy of Leximancer and got busy. I will feed Leximancer various texts over the next few days. I began with a rather hefty challenge: The transcript of Eddie Izzard's Dressed to Kill DVD.

The initial sweep on default settings yielded this (click to enlarge):


Massively increasing the terms (or concepts) yielded this (click to enlarge):

Which is a little clearer if you remove the concepts themselves and just leave the themes (click to enlarge):


And here is some ranking (click to enlarge):


So what can we tell?
  • Eddie has a bit of a foul mouth.
  • Eddie has a "thing" about religion.
  • And war.
  • And cake.
  • Eddie is probably British.
I then unleashed Leximancer on this transcript of Bill Hicks from the early 90s.
Some conclusions for Bill:
  • Like Eddie, Bill also has a foul mouth and issues with religion & food.
  • I really should use this stuff for something more serious.

1 comment:

Barry Saunders said...

hey, nice work, you've picked my two favourite comedians there!