Seamless integration between Tisane API and Coral Project Talk
Tisane API is used as a moderation aid, similar to Perspective API, but with support of 27 languages and more transparency. Tisane allows pinpointing the offending snippet and a type of offense instead of producing a scalar determining "toxicity". Tisane is also much less susceptible to false positives.
Types of abuse detected:
- personal attacks (profane or non-profane)
- hate speech
- sexual advances
- criminal activity
- attempts to establish external contact or arrange payment
See more about the use of Tisane API as a moderation backend here.