A data distribution where a few common categories dominate while many rare categories have few examples.