---

Over the last years, online abuse in different forms has grown from a nuisance to an existential issue.

Some global news websites decided to shut down their comments section: no dialogue is better than a consistently abusive dialogue. Twitter allegedly lost a multi-billion acquisition bid due to the troll problem.

And it’s not just people being nasty with words.

In Singapore, users of illicit drugs locate their dealers using social networks. Australian police is worried about “networked gangs” using social networks to find accomplices. According to CBC, Kik is a haven for pedophiles.

In Germany, UK, and France, various legislative pressures are mounting to make sure the users are not exposed to bigotry, terrorist propaganda, and cyberbullying. UK’s  government is moving towards making the abuse protection mandated by law. The US legislators from both sides of the aisle are concerned about malicious content disseminated via advertisement.

Mark Zuckerberg’s 2018 resolution was to fix abuse in Facebook (Mark, you know where to find us, right?)

 

We are different:

• Tisane API does not merely assign opaque floating point values to find vaguely defined “malicious content”. We pinpoint the offending parts, and classify the abuse by type. We detect bigotry, stated intent to commit crime, sexual advances, personal attacks, and profanities.
• Tisane API detects “subtle” abuse. Existing vendors generally employ a “bag of words” approach. Many of them don’t even account for basics like negations. However, neutral and polite language may very well be used to express hate speech. “Don’t you think the ethnicity X are not as honest as the rest?” “The religion X are not the nicest people” and so on. Just looking up bad words coupled with a group being targeted is like hiring a sniper with poor eyesight.

Read more how to detect abuse with Tisane API.