Reproduce “Estimating Position Bias without Intrusive Interventions” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Aman Agarwal, Ivan Zaitsev, Xuanhui Wang, Cheng Li, Marc Najork, Thorsten Joachims
The authors introduce Intervention Harvesting and AllPairs, which estimate rank-dependent examination propensities from differences in how ranking functions place the same query-result pair, without deliberately swapping results. In Google Drive Search, estimates closely resembled a swap-experiment comparator, unlike regression-EM and raw CTR estimates. This evidence is conditional on the position-based propensity model and comparable ranker assignment.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.