This task involves the work of investigating how we might detect the presence/absence of non-neutral tone in new content edits in a scalable way.
The above will enable, at a minimum, the Editing Team to evaluate to what – if any – extent Tone Check (T365301) impacts the proportion of new content edits (editcheck-newcontent) newcomers and inexperienced volunteers publish that contain non-neutral language.
Absent the specificity this ticket is asking for, we will not be able to conclude that a given Check was effective at causing a particular change in editor behavior beyond the general notion that the edits they publish are less/more likely to be reverted.
Requirements
Identify approaches for aggregating edits that are reverted because they do pose/are at risk of introducing non-neutral language to Wikipedia.
Approaches
- TBD
- TBD
- TBD
Done
- "Approaches" that meet the "Requirements" above are documented that include:
- What tradeoffs we'd need to consider
- A recommendation for which approach to move forward with