Lifestyle

TOPICS:

Previous

PAGE: 

1 ... 

51 | 52 | 53 | 54 | 55

 ... 954

Next

Previous

PAGE: 

1 | 2

Next

motorfirebox

motorfirebox

Pittsburgh, PA
March 2004

JUL 17, 2010 11:17 PM

ElizaTheTroll said:

ElizaTheTroll said:

Shal said:
I have no idea what metric it uses.


My guess is MD5.


Okay, it's not a hash, and it's not just a distribution of word lengths. When I paste in some boilerplate from the gmail home page, I get Arthur Conan Doyle. When I rot13 that, I get Dan Brown (which could be marginally amusing if I were 5 years old).


According to the article linked above, it's mostly about word choice. It's a spam filter, basically, comparing the words you use to the words certain authors use and returning the best match.

ElizaTheTroll

ElizaTheTroll

Australia
January 2006

JUL 18, 2010 12:06 AM

I know what they're claiming it does. It's just that based on its pathetic performance, I suspected they were cheating and using some quasi-random method of picking an author. But now I think they might actually be using some generic analysis program, badly trained with a ridiculously small selection of authors which would appeal to the average internet geek.

...which is probably even lamer than if they had simply cheated.

ElizaTheTroll

ElizaTheTroll

Australia
January 2006

JUL 18, 2010 01:21 AM

So, I found an interview with the author. He openly admits to not having the faintest clue of what he's doing, which I find kind of endearing.

This must be like hitting the jackpot in lotto.

leytonstone

leytonstone

United Kingdom
August 2009

JUL 18, 2010 04:15 AM

ElizaTheTroll said:
I know what they're claiming it does. It's just that based on its pathetic performance, I suspected they were cheating and using some quasi-random method of picking an author. But now I think they might actually be using some generic analysis program, badly trained with a ridiculously small selection of authors which would appeal to the average internet geek.

...which is probably even lamer than if they had simply cheated.



I'll agree with you after using the contents of a couple of reports written in the passive past tense. I write like Dan Brown and Jane Austen - WFT. I Asked my hubby and he said there was no way it could tokenise the strings, parse the content. map the syntactic structure and lots of other stuff I didn't get some of it involving Genetic algorithms. Sad thing is - how many muppets will think their next magnus opus will make then the next Dan Brown. Cluttering up the bookshelves with yet more shite. BTW this comment is in the style of James Joyce

Previous

PAGE: 

1 | 2

Next