Unbiased Learning to Rank

Unbiased learning to rank (ULTR) aims to create accurate ranking models from biased user data, such as click logs, which often overrepresent highly-ranked items. Current research focuses on developing robust algorithms, including those employing inverse propensity weighting (IPW), doubly robust estimation, and contextual dual learning, to mitigate various biases like position bias and contextual bias, often within pairwise or listwise ranking frameworks. These advancements are crucial for improving the performance and fairness of ranking systems in applications like search engines and recommender systems, where biased training data can lead to suboptimal and potentially unfair results.