Modeling Epistemological Principles for Bias Mitigation in AI Systems: An Illustration in Hiring Decisions