Retrieving over corpora that mix several domains often returns relevant but wrong-domain evidence that ranking metrics miss, and conformal risk control bounds only marginally under-cover the worst domains. This work introduces C3R, a drop-in control layer that certifies a per-domain contamination budget from an inferred domain posterior with no query-time label; if infeasible, it abstains rather than silently violating. On the hardest domains, it guarantees a reduction, not a tight bound. The core is a two-split scheme built on risk-controlling prediction sets, whose finite-sample transfer bound crosses from the inferred to the true domain with fully estimable slack, supports heterogeneous budgets, and inverts for deployment. Population validity rests on this bound and a controlled simulation; across a thousand resampled calibrations, the certificate never violates (a stability result) while marginal control violates the most-contaminated domain in every draw, and soft demotion retains more recall than the strongest calibrated cascade at equal certified contamination. The method replicates across open testbeds including an independent one from public federal regulations, and an LLM-judged downstream probe indicates wrong-authority grounding rises with contamination and falls under control. The layer is frozen-stack and reranker-agnostic.
Blogger's Review: The innovative aspect of the C3R control layer lies in its label-free contamination control mechanism, providing a more reliable multi-domain retrieval capability. Through effective risk control and rigorous experimental validation, it demonstrates potential in real-world applications, making it worthy of attention and exploration.