Add-Link-Structured-Task suggestions sometimes prompt erroneous links of proper noun homonyms (particularly people and places), such as to a [[Jane Smith]] who was a writer when the actual link was referencing a Jane Smith who was a swimmer. Newcomers often do not recognize that the entities referenced are different, and add an erroneous link as a result.

This task captures the work of reducing the prevalence of this phenomenon. Potential strategies include:

• Education in the onboarding or interface to help make newcomers aware of this risk.
• Machine learning improvements to make more context-aware link suggestions more likely to go to the right entity
• Machine learning improvements to help the model understand which links are most likely to be homonyms, and either not suggesting these or flagging them for newcomers as requiring extra caution (this could perhaps involve looking at the presence of hatnotes or linked disambiguation pages)
• Community Configuration settings to enable editors to turn off suggestions to types of entities where the phenomenon is particularly bad

This task was inspired by feedback from English Wikipedia, primarily here, here, and here.