We develop an algorithm to train individually fair learning-to-rank (LTR)
models. The proposed approach ensures items from minority groups appear
alongside similar items from majority groups. This notion of fair ranking is
based on the definition of individual fairness from supervised