The True Knowledge Profanity Filter

Recently we opened our site to user-added free-text answers.  There's
been a lot of content added - mostly great, but not surprisingly some
has been unsuitable for display.

We experimented with rejecting answers using an off-the-shelf profanity
filter but weren't too impressed with the results.  Some profanities
were being allowed through and the false positives upset our innocent
users.  Answers were being rejected if they were about 'Hellenism',
'assertions' or the answer to Who is Sunday's child?

So we built our own.  It's proved successful to date and having access to our massive ontology has even helped us solve the Scunthorpe Problem.

We'd like to share this service with our API users.  You'll find it useful if your site

• displays comments or adverts from the public
• features messaging between users (particularly minors)
• aggregates third party content
• allows people to choose their own usernames

or anything else involving free text.

You can ask our API to find whether a string is likely to be profane or not simply by running a query like this

query
[string is profane] [applies to] ["assertions about hellenism"]

Sign up to our Query Service API to use this plus all our other query services.