Rank discrepancies expose the rift between training data associations and live search reality by comparing mean ordinal placements and rank stability metrics across paired ungrounded and grounded runs.
Referenced by
