15 comments

[ 1.9 ms ] story [ 47.5 ms ] thread
Interesting, which libraries did you use?
Natural Language Toolkit (http://www.nltk.org/) for Python.

The initial code for the "insulter.py", the core, was no more than 150 lines. We'll probably make our repository public at some point so you can check it out.

Are the thumbs up/down used to train the system? Looking forward to the code
Yes. In a very primitive way, at the moment, just updating the weights for adjectives
Pretty neat. From solely using that the prepopulated example input, my main gripe would be that it doesn't get a/an correct in the "cursed up" version of the sentence.
Funny we didn't notice that!

Yes, that's true. And that's a fairly poorly written example.

This reminds me of the old dialectizer that would rewrite web pages in specific dialects.
this comes nowhere-the-fuck near cursing as dynamically as i can. ;)
Gives me an idea for a new kind of Turing test ...
It's a neat tool but I wish it could better parse paragraphs and commas. I also had a few responses with the same curse words three or four times in a block of text making it a fucking boring read.
Yep, our dictionaries for adjectives, adverbs and nouns are tiny at the moment
I was thinking it would be more like readability, but instead of reformatting the html, inserting curse words. How much more fun would cnn.com be?! But a good start!