Re-Rankers as Relevance Judges Conference: Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval; July 13–18, 2025; Padua, Italy Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval (
Chuan Meng, Jiqun Liu, Mohammad Aliannejadi, Fengran Mo, Jeff Dalton, Maarten de Rijke · 2026
Read Re-Rankers as Relevance Judges Conference: Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval; July 13–18, 2025; Padua, Italy Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval ( by Chuan Meng, Jiqun Liu, Mohammad Aliannejadi, Fengran Mo, Jeff Dalton, Maarten de Rijke (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-04455.txt.
Description
Chuan Meng and five co-authors ask whether the tools that rank search results can also judge them, in a 2026 arXiv paper. Relevance judgments — human-style labels of whether a passage answers a query — are increasingly predicted by large language models at great expense. Yet re-rankers, built to order results by relevance, are solving a near-identical problem, and reusing them could save duplicated effort.
The team adapts eight re-rankers from three families two ways: reading their generated true-or-false tokens as verdicts, or thresholding their scores into labels. On five years of TREC deep learning track data, the repurposed judges beat a leading language-model judge in around half the cases. They also caught a bias worth knowing: re-ranker judges favour results from their own and related re-rankers, a self-preference that any evaluation built on them must correct for.
Catalog record
| CREATOR | Chuan Meng, Jiqun Liu, Mohammad Aliannejadi, Fengran Mo, Jeff Dalton, Maarten de Rijke |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-04455