Wikipedia:Bots/Requests for approval/DetoxAgent


 * The following discussion is an archived debate. Please do not modify it. To request review of this BRFA, please start a new section at Bots/Noticeboard. The result of the discussion was

DetoxAgent
Operator: Alphabet Jigsaw

Time filed: 19:07, Tuesday, April 2, 2019 (UTC)

Function overview: DetoxAgent is a machine learning based assistant that identifies offensive or insulting comments when they appear on article Talk pages. The agent ranks comments and provides users a simple interface to see the comments in context, and perform annotated reverts.

Automatic, Supervised, or Manual: Manual

Programming language(s): Python

Source code available: https://github.com/conversationai/wikidetox

Links to relevant discussions (where appropriate):

Edit period(s): Continuous

Estimated number of pages affected:

Namespace(s): Mainspace/Articles:Talk

Exclusion compliant (Yes/No):

Function details: Jigsaw has been developing machine learning models for detecting "Toxic" comments, where toxic is defined to be comments that are rude, offensive, and likely to make someone want to leave the conversation. These models have been trained on supervised data created using crowd workers. This work has been published documented in recent conferences.

We intend to create a website that displays graphically the current talk page comments that we detect having these attributes, and will allow visitors to review the comments, see them in context, and perform reverts, if desired, on a case by case basis.

Discussion
— Preceding unsigned comment added by PaigePhault (talk • contribs) 20:19, 27 April 2021 (UTC)
 * The above discussion is preserved as an archive of the debate. Please do not modify it. To request review of this BRFA, please start a new section at Bots/Noticeboard.