![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Gender Genie
Apr. 20th, 2008 11:59 amSome people were chatting about the "Gender Genie" on the SFReader forum today and yesterday, and whether the genie could 'predict' what would get chosen for Sword and Sorceress 23.
So I ran a couple of my stories through, just to see what they got.
Story #1...Male Author
Story #2...Female Author (by a large margin)
I think what most interested me was the algorithm apparently used to judge the text. Apparently the algorithm searches for specific words identified as male or female, counts the frequency, and then assigns a weight to each word.
For example, every single with gets 52 estrogen points, while each around gets 42 testosterone points. Multply all the estrogen words by their frequency and you get your 'femaleness'. Ditto the testosterone words.
I suppose the most interesting part of this is the words that seem to fall into the 'estrogen' category and their specific weighting, versus the testosterone words.
So, let's consider the BIG estrogen words (in order of importance)
With--52 points per use; if--47 ppu, not--27 ppu; where--18 ppu; be, when, and your--17 ppu.
The biggest guy words?
Around--42, what--35, more--34, are--28, as--23, who--19
Some interesting facts I noticed:
1) The women are asking where, but the men aren't. See, men really don't ask for directions!
2) Pronouns ended up on the estrogen count. Your, her, we, she, me, myself, and hers all made the estrogen list. Not a single pronoun on the men's side--except IT. I don't want to draw conclusions from that.
3) Women had no prepositions. Men had around, below, above, and to. For some reason, men also got the articles the and a. I wonder how male Russian writers fare on this test...
Anyway, the test gave me food for thought....
And this blog entry?
Female Score: 494
Male Score: 678
So I ran a couple of my stories through, just to see what they got.
Story #1...Male Author
Story #2...Female Author (by a large margin)
I think what most interested me was the algorithm apparently used to judge the text. Apparently the algorithm searches for specific words identified as male or female, counts the frequency, and then assigns a weight to each word.
For example, every single with gets 52 estrogen points, while each around gets 42 testosterone points. Multply all the estrogen words by their frequency and you get your 'femaleness'. Ditto the testosterone words.
I suppose the most interesting part of this is the words that seem to fall into the 'estrogen' category and their specific weighting, versus the testosterone words.
So, let's consider the BIG estrogen words (in order of importance)
With--52 points per use; if--47 ppu, not--27 ppu; where--18 ppu; be, when, and your--17 ppu.
The biggest guy words?
Around--42, what--35, more--34, are--28, as--23, who--19
Some interesting facts I noticed:
1) The women are asking where, but the men aren't. See, men really don't ask for directions!
2) Pronouns ended up on the estrogen count. Your, her, we, she, me, myself, and hers all made the estrogen list. Not a single pronoun on the men's side--except IT. I don't want to draw conclusions from that.
3) Women had no prepositions. Men had around, below, above, and to. For some reason, men also got the articles the and a. I wonder how male Russian writers fare on this test...
Anyway, the test gave me food for thought....
And this blog entry?
Female Score: 494
Male Score: 678
The Gender Genie thinks the author of this passage is: male!
Just goes to show you, GG is not all-knowing!
