---

The detected abuse is included in the response if the abuse setting is set to true. It is the default value, so essentially, it is shown by default, unless it is explicitly set to false.

All the instances of abuse are stored in the abuse block of the response. Every instance of abuse has the following attributes:

• type – a type of the abuse. Currently we detect these kinds of abuse:
  • bigotry – bigotry toward an ethnic or a religious group or a minority or a gender
  • personal_attack – a personal attack on a participant of the online social circle (including cyberbullying)
  • criminal_activity – planning criminal activity
  • sexual_advances – unsolicited sexual advances
  • profanity – bad language
  • external_contact – attempt to establish contact or obtain payment outside of current media (e.g. by phone or a messaging application)
  • generic – reserved
• severity_level – how severe the abuse is. The current gradation is:
  • low – the lowest level of abuse
  • medium – abuse of medium severity
  • high – severe abuse
  • extreme – extremely severe abuse

Additional information allows obtaining the text fragment with the instance of abuse. See How do I locate a snippet based on the location data? to learn how to obtain the text fragment.

Refer to the documentation on various types of abuse for more type-specific information.

Also see How do I use the abuse detection functionality with live user content.