Annotated bibliography and synthesis of process frameworks for bias mitigation in algorithmic systems. Draw out guidance that is most relevant to our products, policies and practices for review with stakeholders.
|Resolved||• Capt_Swing||T203048 Output 1.0: A unit test to measure the bias of recommendation algorithms|
|Resolved||• Capt_Swing||T206051 Iteration on recommender system bias and auditability report|
|Resolved||• Capt_Swing||T207513 Synthesize algorithmic bias mitigation strategies from existing process frameworks|
Process frameworks are glossed here: https://meta.wikimedia.org/wiki/Research:Ethical_and_human-centered_AI/Process_frameworks
Information from those frameworks has been integrated into the process proposal section of the on-wiki report here: https://meta.wikimedia.org/wiki/Research:Ethical_and_human-centered_AI#Process_proposals and will be summarized in the white paper.