This plugin implements ModerateHatespeech.com’s machine learning API to detect and flag potentially toxic and hateful comments. A configurable threshold can be set and comments with scores above that threshold will be sent to the moderation queue for administrative approval.
A free API key from ModerateHatespeech.com is required.
Defining Toxic
We define “Toxic” as “any content a reasonable, PG-13 forum wouldn’t want for content reasons.”
This does not include:
- Spam or Advertisements
- NSFW/Explicit Content
- Forum-specific Etiquette
This does include:
- Hate Speech
- Verbal Insults
- Slurs Directed Towards a Group
More details on toxicity, as well as model accuracy and bias can be found on ModerateHatespeech.com
This plugin is part of the ModerateHatespeech project, an initiative to research and reduce online hate. Use of the API constitues acceptance of ModerateHatespeech’s Terms of Service and Privacy Policies.