Recommendation Ranking Off-Policy Evaluation under Ranking-Dependent Examination via Examination-Relevance Decomposition
Two estimators based on the decomposition of clicks into examination and relevance are proposed, one of which corrects the IIPS bias using policy examination probability ratios and extends LE-IIPS to a doubly robust framework.
Riki Okamura, Toshiharu Sugawara
· 0 citations