Sep. 23rd, 2003
![[livejournal.com profile]](https://www.dreamwidth.org/img/external/lj-userinfo.gif){kind=small}
Strange... so strange
Sep. 23rd, 2003 10:30 amOkay, I may be a little late getting the news, but some wacky guys named Koppel and Argamon have devised an algorithm that checks fiction for certain words and basically decides if the author is male or female.
The algorithm is absurdly simplistic. You get 'male' points for the following words: the, a, some, [numbers], it
You get 'female' points for: with, [possessives], for, not
The idea seems ludicrous, but apparently there's some merit to it. It's simple enough that it has been turned into tools on the web, where you can cut and paste into the text box and analyze away. And now there's a tool for analyzing LiveJournals.
Here are some results picked off my Friends list:
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
essentialsaltes - 70% male
chibi_neko - 63% male [wrong]
bellwethr - 60% male
talentlessclod - 60% male
mersh - 59% male
britgeekgrrl - 54% male [wrong]
mayxbo5 - 49% male
tardpenguin - 49% male
dreamguy - 46% male [wrong]
oohbarracuda - 46% male
n0t2shabby - 45% male
It's interesting that the females are mostly clustered from 45-49, so maybe the scale should be shifted a little bit. I.e., if the dividing line were between mersh and britgeekgrrl, the test would have one more correct answer, and girly-girls like Ice and Elke would be further from from the dividing line. Of course, since my writing is apparently so powerfully rugged and masculine, I wouldn't mind if the scale were shifted 5%, but maybe there's some limp-wristed pansies out there who would lispingly complain.
Time to go read some of Two-Gun Bob Howard's stuff.
The algorithm is absurdly simplistic. You get 'male' points for the following words: the, a, some, [numbers], it
You get 'female' points for: with, [possessives], for, not
The idea seems ludicrous, but apparently there's some merit to it. It's simple enough that it has been turned into tools on the web, where you can cut and paste into the text box and analyze away. And now there's a tool for analyzing LiveJournals.
Here are some results picked off my Friends list:
essentialsaltes - 70% malechibi_neko - 63% male [wrong]bellwethr - 60% maletalentlessclod - 60% malemersh - 59% malebritgeekgrrl - 54% male [wrong]mayxbo5 - 49% maletardpenguin - 49% maledreamguy - 46% male [wrong]oohbarracuda - 46% malen0t2shabby - 45% maleIt's interesting that the females are mostly clustered from 45-49, so maybe the scale should be shifted a little bit. I.e., if the dividing line were between mersh and britgeekgrrl, the test would have one more correct answer, and girly-girls like Ice and Elke would be further from from the dividing line. Of course, since my writing is apparently so powerfully rugged and masculine, I wouldn't mind if the scale were shifted 5%, but maybe there's some limp-wristed pansies out there who would lispingly complain.
Time to go read some of Two-Gun Bob Howard's stuff.