₦airaland Forum

THE Algorithm for capturing banned words for NL is so easy to bypass.
this is just a simple algorithm that can catch certain cases.
An all case scenario filter would i think be very complex.

catches cases like ban word , punctuated with non alphabetic characters.
e.g
if banned word = ban,banned,banner.
and user puts ba.n, ban.ned, etc
.
algorithm:
------------
one- .
split sentence into words.
-
- remove non alphabetic characters from words.
- check word against banned word database,
- if match, substitute banned word with NL version.

e.g ba.n -> ban (after removing non alphabetic characters -> check against NL db . match found found -.
replace.

fu....ck, b.rea.st, etc... all in the net now.

oga catch this ones:

breeast
fucck
boobi
pusi
pussyy

etc.

truth is, no matter how smart these bots are, humans will get smarter. Afterall they were designed by humans.

Nice post though.

sarutobi: oga catch this ones:

breeast
fucck
boobi
pusi
pussyy

etc.

truth is, no matter how smart these bots are, humans will get smarter. Afterall they were designed by humans.

Nice post though.

This would need a complex regex

sarutobi: oga catch this ones:

breeast
fucck
boobi
pusi
pussyy

etc.

truth is, no matter how smart these bots are, humans will get smarter. Afterall they were designed by humans.

Nice post though.

U b baaad guy
I thnk dis will need a whole english dico to be put in the database, every word u typed is checked against d database, if not found, it is substituted with d nearest one and chcked against d banned words, if it matches d banned word, then an action is taken!

hahaha @sarutobi... funny examples...

Though, the best way to build up algorithms to take care of cases like that is to use: Nearest Neighbor heuristics; and high collision probability LSH L(ocality-sensitive hashing)....

Well, banned words system is either a lazy or a simple forum filtering system . . . for a revenue generating website like NL...
How about banned words + phrases....?? Grammar and linguistic systems. . . start implementing primitive systems as such.... its hard but good for commercial stuff....

Seun has a lot of big data of comments on NL, we could build such Probabilistic filtering systems.... And make it learn the kind of grammar it should ban... from a set of comments with insults....