Javascript AntiSwear Leet

herohamp

Does anyone know a way to make something detect for swear words and leet versions of them? like for instance lets say “Cake” is a bad word how would I detect if that is in a string and also detect if C@ke C@k3 Cak3 is in a string without having to type every single leet alternative into an array of bad works

cheeseeater

Translate ‘leet’ characters back to their English alphabet counterparts and then check the inappropriate word list.

herohamp

cheeseeater wrote:
Translate ‘leet’ characters back to their English alphabet counterparts and then check the inappropriate word list.

ok

comp09

My advice: implement a basic regex generator, and give up. Nothing beats humans.

herohamp

comp09 wrote:
My advice: implement a basic regex generator, and give up. Nothing beats humans.

done already

MegaApuTurkUltra

comp09 wrote:
My advice: implement a basic regex generator, and give up. Nothing beats humans.

^^^

iamunknown2

comp09 wrote:
Nothing beats humans.

What if you trained a neural network to recognize swear words?

nanalan

iamunknown2 wrote:
comp09 wrote:
Nothing beats humans.
What if you trained a neural network to recognize swear words?

comp09 wrote:
Nothing beats humans.

BookOwl

nanalan wrote:
iamunknown2 wrote:
comp09 wrote:
Nothing beats humans.
What if you trained a neural network to recognize swear words?
comp09 wrote:
Nothing beats humans.

Yet.

comp09

See: Tom Scott's video about the Scunthorpe problem.

Don't make the clubuttic mistake of inventing the Donkeyrians.

Last edited by comp09 (Aug. 15, 2016 13:19:22)

iamunknown2

nanalan wrote:
iamunknown2 wrote:
comp09 wrote:
Nothing beats humans.
What if you trained a neural network to recognize swear words?
comp09 wrote:
Nothing beats humans.

Say that to Go masters

Last edited by iamunknown2 (Aug. 15, 2016 13:33:42)

herohamp

comp09 wrote:
See: Tom Scott's video about the Scunthorpe problem.

Don't make the clubuttic mistake of inventing the Donkeyrians.

welp I hope nobody names there self one of those names then ;P

technoboy10

It might also be interesting to implement some sort of string distance calculator to prevent this type of thing (“cake” and “c@k3” have a Levenshtein distance of 2, which might fall into some threshold for detection).

comp09

technoboy10 wrote:
It might also be interesting to implement some sort of string distance calculator to prevent this type of thing (“cake” and “c@k3” have a Levenshtein distance of 2, which might fall into some threshold for detection).

And then you get CleanSpeak filtering “FCC” out of Scratch comments…

kvackkvack

iamunknown2 wrote:
nanalan wrote:
iamunknown2 wrote:
comp09 wrote:
Nothing beats humans.
What if you trained a neural network to recognize swear words?
comp09 wrote:
Nothing beats humans.
Say that to Go masters

Recognising human intentions through written text =/= mastering Go

Last edited by kvackkvack (Aug. 15, 2016 17:16:28)

Pturretdactyl

comp09 wrote:
technoboy10 wrote:
It might also be interesting to implement some sort of string distance calculator to prevent this type of thing (“cake” and “c@k3” have a Levenshtein distance of 2, which might fall into some threshold for detection).
And then you get CleanSpeak filtering “FCC” out of Scratch comments…

I've had ‘Cartron’ blocked several times, and once I had ‘11:30ish’ blocked. My joke is that the world will end one day at 11:30ish.

kvackkvack wrote:
iamunknown2 wrote:
nanalan wrote:
iamunknown2 wrote:
comp09 wrote:
Nothing beats humans.
What if you trained a neural network to recognize swear words?
comp09 wrote:
Nothing beats humans.
Say that to Go masters
Recognising human intentions through written text =/= mastering Go

Point is that something beat a human.

Just a thought… What if there was a competition based on this? It would be something like you have to develop a program that blocks out certain words, and there is a set of users who are trying to get past the filter somehow. Through several stages, you refine your code based on what the ‘human group’ does to get around the blocks.

And the winner gets the filter implemented in IBM's Watson!

herohamp

or maybe I can just get all the boys in my grade to try and swear on my chat then have my system learn

Firedrake969

Pturretdactyl wrote:
comp09 wrote:
technoboy10 wrote:
It might also be interesting to implement some sort of string distance calculator to prevent this type of thing (“cake” and “c@k3” have a Levenshtein distance of 2, which might fall into some threshold for detection).
And then you get CleanSpeak filtering “FCC” out of Scratch comments…
I've had ‘Cartron’ blocked several times, and once I had ‘11:30ish’ blocked. My joke is that the world will end one day at 11:30ish.
kvackkvack wrote:
iamunknown2 wrote:
nanalan wrote:
iamunknown2 wrote:
comp09 wrote:
Nothing beats humans.
What if you trained a neural network to recognize swear words?
comp09 wrote:
Nothing beats humans.
Say that to Go masters
Recognising human intentions through written text =/= mastering Go
Point is that something beat a human.

Just a thought… What if there was a competition based on this? It would be something like you have to develop a program that blocks out certain words, and there is a set of users who are trying to get past the filter somehow. Through several stages, you refine your code based on what the ‘human group’ does to get around the blocks.

And the winner gets the filter implemented in IBM's Watson!

That's called capitalism

Pturretdactyl

Firedrake969 wrote:
Pturretdactyl wrote:
-snip-
Point is that something beat a human.

Just a thought… What if there was a competition based on this? It would be something like you have to develop a program that blocks out certain words, and there is a set of users who are trying to get past the filter somehow. Through several stages, you refine your code based on what the ‘human group’ does to get around the blocks.

And the winner gets the filter implemented in IBM's Watson!
That's called capitalism

'Course if I had the winning program, I'd stealthily add in some… interesting… substitutions. Scientist = wizard, dictionary = Pokedex, etc.

The Munroe Machine.

Last edited by Pturretdactyl (Aug. 15, 2016 22:34:08)

Jonathan50

comp09 wrote:
technoboy10 wrote:
It might also be interesting to implement some sort of string distance calculator to prevent this type of thing (“cake” and “c@k3” have a Levenshtein distance of 2, which might fall into some threshold for detection).
And then you get CleanSpeak filtering “FCC” out of Scratch comments…

Why does that happen?

Discuss Scratch