arxiv: 2604.27577 · v1 · submitted 2026-04-30 · 💻 cs.IR

Recognition: unknown

Reproducing Adaptive Reranking for Reasoning-Intensive IR

Mandeep Rathee , V Venktesh , Sean MacAvaney , Avishek Anand

Authors on Pith no claims yet

Pith reviewed 2026-05-07 08:01 UTC · model grok-4.3

classification 💻 cs.IR

keywords information retrievalrerankinggraph-based adaptive rerankingreasoning-intensive queriesbounded recallreplicationBRIGHT benchmark

0 comments

The pith

Graph-based adaptive reranking boosts reasoning-intensive retrieval with minimal overhead

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper replicates Graph-based Adaptive Reranking (GAR) on the BRIGHT benchmark for reasoning-intensive information retrieval. GAR modifies the reranking stage to iteratively explore a corpus graph and recover relevant documents missed by the first-stage retriever. The replication tests both reasoning and non-reasoning rerankers and finds that the strength of the reranker's signal determines success in the graph exploration. A sympathetic reader cares because the method delivers effectiveness gains without the substantial costs of retraining or replacing the initial retriever for complex queries.

Core claim

GAR addresses the bounded recall problem by modifying the reranking process itself through iterative exploration of a corpus graph. Replicated on the BRIGHT reasoning-intensive retrieval benchmark, GAR boosts the effectiveness of retrieval across a variety of models while contributing minimally to computational overheads. The quality of the reranker's signal plays an important role in identifying additional relevant documents within the corpus graph.

What carries the argument

Graph-based Adaptive Reranking (GAR), a reranking technique that uses iterative traversal of a corpus graph guided by the reranker's relevance scores to expand the candidate set beyond the initial retrieval results.

If this is right

GAR improves retrieval metrics on the BRIGHT benchmark for reasoning-intensive queries.
The improvements apply to both reasoning and non-reasoning reranking models.
The method adds only minimal computational overhead compared to standard reranking.
It serves as a low-cost way to mitigate bounded recall without enhancing the first-stage retriever.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The findings imply that investing in higher-quality rerankers may yield compounding benefits when combined with graph-based adaptation.
This approach could be extended to other retrieval benchmarks involving multi-step or complex reasoning queries.
In production systems, GAR offers a modular addition that improves handling of diverse query difficulties without full pipeline retraining.

Load-bearing premise

The reranker's relevance scores remain reliable enough to direct productive exploration through the corpus graph even for queries that require substantial reasoning.

What would settle it

A direct replication experiment on BRIGHT showing no statistically significant gains in effectiveness metrics like nDCG or recall when GAR is applied, or a measured inference time increase that exceeds the minimal overhead reported.

Figures

Figures reproduced from arXiv: 2604.27577 by Avishek Anand, Mandeep Rathee, Sean MacAvaney, V Venktesh.

**Figure 1.** Figure 1: Overview of Graph-based Adaptive Reranking. The view at source ↗

**Figure 2.** Figure 2: Qualitative example for showing the position of view at source ↗

**Figure 3.** Figure 3: Retrieval performance in Recall@100 across different methods on the StackExchange subset of BRIGHT. First, BM25 view at source ↗

**Figure 4.** Figure 4: Retrieval and ranking performance of Gar when augmented with different re-rankers and varying batch size 𝑏. The first row shows nDCG@10 and second shows Recall@50 performance. showed that Gar is robust to the selection of these hyperparameters. We reproduce Gar by varying batch size 𝑏 and neighbors 𝑘. In the first experiment, we fix the number of neighbors 𝑘 to 16 and vary 𝑏 ∈ [2, 4, 8, 16, 32]. We do not … view at source ↗

**Figure 5.** Figure 5: Effect of number of neighbors in the corpus graph. The first row shows nDCG@10 performance, and the second row view at source ↗

**Figure 6.** Figure 6: We observe that Gar remain highly effective at a lower budget as well. For instance, TFRank-0.6B with Gar improves average nDCG@10 from 16.5 to 18.8 and TFRank-4B from 21.7 to 24. More interestingly, we observe that the performance of the reranking pipeline, which reranks 100 documents, can be achieved by reranking just 50 documents (0.5x), which reduces the computational cost, latency, and carbon footpr… view at source ↗

**Figure 7.** Figure 7: Performance comparison (avg. nDCG@10) on view at source ↗

read the original abstract

The classical cascading pipeline of retrieve--rerank suffers from a bounded recall problem, stemming from limitations of the first-stage retriever. Most current approaches address the bounded recall problem by improving the first-stage retriever, but this incurs substantial training and inference costs, especially to handle queries that require substantial reasoning. To circumvent the computational costs of reasoning-based retrievers, we replicate the findings of GAR, Graph-based Adaptive Reranking, on the BRIGHT reasoning-intensive retrieval benchmark. GAR addresses the bounded recall problem by modifying the reranking process itself through iterative exploration of a corpus graph, but it was previously only tested on models designed for topical and question-answering-style queries. Hence, reproduce GAR in reasoning-intensive settings with reasoning and non-reasoning reranking models. We observe that the quality of the reranker's signal plays an important role in identifying additional relevant documents within the corpus graph. Overall, we find that GAR boosts the effectiveness of reasoning-intensive retrieval across a variety of models while contributing minimally to computational overheads. Ultimately, this work enables more practical deployment of retrieval systems that can address reasoning-intensive queries.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This is a useful reproduction of GAR on BRIGHT, but the key observation on reranker quality lacks supporting ablations.

read the letter

The main thing to know is that this paper reproduces the GAR method on the BRIGHT benchmark and finds it improves retrieval for reasoning-intensive queries with minimal added cost. It also notes that reranker signal quality seems important for the graph-based gains. They apply the iterative corpus graph exploration from GAR to a new set of reasoning-heavy queries. The work tests a range of reranking models, including ones designed for reasoning. This shows the approach can transfer beyond the original topical and QA settings, which is a solid check on generalizability. The low computational overhead is a practical plus for avoiding expensive retriever retraining. The soft spots are in the evidence for the mechanism. The abstract claims the reranker's signal plays an important role, but there's no ablation that varies reranker quality while keeping the graph exploration fixed, or vice versa. No per-query analysis or low-quality reranker control is mentioned. In reasoning settings, where relevance can hinge on multi-hop logic rather than direct matches, it's possible the gains come from other factors like the initial retrieval or benchmark specifics. Without those checks, the observation stays at the level of a note rather than a tested finding. The paper also doesn't provide any quantitative results, error bars, or statistical details in the summary, which makes it tough to gauge the size of the improvements. This is aimed at IR researchers and engineers working on systems for complex, reasoning-based search tasks, such as in legal, academic, or advanced QA applications. Readers interested in efficient reranking techniques or validation of existing methods on new benchmarks will get the most out of it. It deserves a serious referee because reproductions that extend methods to challenging new domains like BRIGHT help the field, even if they require more detail. I'd recommend sending it to peer review, with the expectation that the authors add the ablations and full results to strengthen the central claims.

Referee Report

2 major / 2 minor

Summary. The paper is a reproduction study of Graph-based Adaptive Reranking (GAR) applied to the BRIGHT benchmark, which focuses on reasoning-intensive information retrieval. The authors claim that GAR successfully addresses the bounded recall issue in retrieve-rerank pipelines for such queries by iteratively exploring a corpus graph, leading to improved effectiveness across different reranking models with only minimal added computational overhead. They further observe that the reranker's signal quality is key to the success of this graph-based exploration in surfacing additional relevant documents.

Significance. Should the reproduction be confirmed with rigorous ablations and quantitative evidence, the findings would hold moderate significance for the IR community. They suggest a cost-effective alternative to developing specialized reasoning retrievers, potentially broadening access to high-performance retrieval for complex queries in applications like scientific literature search or legal document retrieval. The work also highlights the transferability of GAR to new query types.

major comments (2)

[Abstract] The central observation that reranker signal quality plays an important role lacks any reported ablation, correlation analysis, or control experiment on the BRIGHT dataset. Without testing a low-quality reranker variant or measuring signal strength vs. gain, the mechanistic claim remains unverified and could be confounded by first-stage retriever properties or dataset specifics.
[Results] The abstract provides no quantitative results, error bars, ablation details, or statistical tests to support the claim that GAR boosts effectiveness across models. Full details on metrics, improvements, and variability are required to evaluate the practical impact and robustness.

minor comments (2)

[Methods] Provide more details on the specific reasoning and non-reasoning reranking models used, as well as the exact configuration of the corpus graph construction and iteration parameters for GAR.
Consider adding a table summarizing the main results with baseline comparisons, including standard deviations if multiple runs were performed.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments on our reproduction study. We address each major comment below and will incorporate revisions to improve the clarity and rigor of the manuscript, particularly in the abstract and supporting analyses.

read point-by-point responses

Referee: [Abstract] The central observation that reranker signal quality plays an important role lacks any reported ablation, correlation analysis, or control experiment on the BRIGHT dataset. Without testing a low-quality reranker variant or measuring signal strength vs. gain, the mechanistic claim remains unverified and could be confounded by first-stage retriever properties or dataset specifics.

Authors: We agree that a more explicit analysis would strengthen the mechanistic claim. Our experiments do compare GAR performance across rerankers with varying reasoning capabilities on BRIGHT, which provides indirect evidence for the importance of signal quality. However, we did not include a dedicated correlation analysis or control with a deliberately low-quality reranker variant. In the revision, we will add a targeted analysis (e.g., correlating initial reranker effectiveness with GAR-induced gains) and, if feasible, results from a weakened reranker signal to help rule out confounding factors from the first-stage retriever or dataset. revision: yes
Referee: [Results] The abstract provides no quantitative results, error bars, ablation details, or statistical tests to support the claim that GAR boosts effectiveness across models. Full details on metrics, improvements, and variability are required to evaluate the practical impact and robustness.

Authors: We agree that the abstract would benefit from including key quantitative summaries. The full manuscript already reports metrics (e.g., nDCG and recall improvements), model-specific results, and computational overhead comparisons across rerankers. In the revision, we will update the abstract to include representative quantitative findings (such as average effectiveness gains), reference the presence of variability measures in the results, and ensure that error bars, ablation details, and any statistical tests are clearly highlighted in the main text and figures. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical reproduction of external method on new benchmark

full rationale

The paper is a reproduction study applying the existing GAR method to the BRIGHT reasoning-intensive benchmark. No equations, derivations, fitted parameters, or predictions are introduced. All claims rest on empirical effectiveness metrics (e.g., retrieval performance across reranking models) rather than any self-referential construction. References to the original GAR work are external citations for the method being tested, not load-bearing self-citations or ansatzes smuggled in. The observation that reranker signal quality matters is presented as an empirical finding from the experiments, not a definitional or fitted tautology. The study is therefore self-contained against external benchmarks with no reduction of results to inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No mathematical derivations, free parameters, axioms, or invented entities are present in the abstract; the work is purely empirical.

pith-pipeline@v0.9.0 · 5500 in / 1001 out tokens · 43221 ms · 2026-05-07T08:01:57.861219+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

54 extracted references · 33 canonical work pages · 4 internal anchors

[1]

Seonho An, Chaejeong Hyun, and Min-Soo Kim. 2026. FastInsight: Fast and Insightful Retrieval via Fusion Operators for Graph RAG.arXiv preprint arXiv:2601.18579(2026)

work page arXiv 2026
[2]

Akari Asai, Timo Schick, Patrick Lewis, Xilun Chen, Gautier Izacard, Sebastian Riedel, Hannaneh Hajishirzi, and Wen-tau Yih. 2023. Task-aware Retrieval with Instructions. InFindings of the Association for Computational Linguistics: ACL 2023, Anna Rogers, Jordan Boyd-Graber, and Naoaki Okazaki (Eds.). Association for Computational Linguistics, Toronto, Can...

work page doi:10.18653/v1/ 2023
[3]

Sebastian Bruch, Siyu Gai, and Amir Ingber. 2023. An analysis of fusion functions for hybrid retrieval.ACM Transactions on Information Systems42, 1 (2023), 1–35

2023
[4]

Tao Chen, Mingyang Zhang, Jing Lu, Michael Bendersky, and Marc Najork. 2022. Out-of-domain semantics to the rescue! zero-shot hybrid retrieval models. In European Conference on Information Retrieval. Springer, 95–110

2022
[5]

Zijian Chen, Xueguang Ma, Shengyao Zhuang, Ping Nie, Kai Zou, Sahel Shar- ifymoghaddam, Andrew Liu, Joshua Green, Kshama Patel, Ruoxi Meng, et al
[6]

InFirst Workshop on Multi-Turn Interactions in Large Language Models

BrowseComp-Plus: A More Fair and Transparent Evaluation Benchmark of Deep-Research Agent. InFirst Workshop on Multi-Turn Interactions in Large Language Models
[7]

Cormack, Charles L A Clarke, and Stefan Buettcher

Gordon V. Cormack, Charles L A Clarke, and Stefan Buettcher. 2009. Reciprocal rank fusion outperforms condorcet and individual rank learning methods. In Proceedings of the 32nd International ACM SIGIR Conference on Research and Development in Information Retrieval(Boston, MA, USA)(SIGIR ’09). Association for Computing Machinery, New York, NY, USA, 758–759...

work page doi:10.1145/1571941 2009
[8]

Lachlan Dunn, Luke Gallagher, and Joel Mackenzie. 2025. Approximate Bag- of-Words Top-k Corpus Graphs. InAdvances in Information Retrieval, Claudia Hauff, Craig Macdonald, Dietmar Jannach, Gabriella Kazai, Franco Maria Nardini, Fabio Pinelli, Fabrizio Silvestri, and Nicola Tonellotto (Eds.). Springer Nature Switzerland, Cham, 174–182

2025
[9]

Darren Edge, Ha Trinh, Newman Cheng, Joshua Bradley, Alex Chao, Apurva Mody, Steven Truitt, Dasha Metropolitansky, Robert Osazuwa Ness, and Jonathan Larson. 2024. From local to global: A graph rag approach to query-focused summarization.arXiv preprint arXiv:2404.16130(2024)

work page internal anchor Pith review arXiv 2024
[10]

Yongqi Fan, Xiaoyang Chen, Dezhi Ye, Jie Liu, Haijin Liang, Jin Ma, Ben He, Yingfei Sun, and Tong Ruan. 2025. TFRank: Think-Free Reasoning Enables Practical Pointwise LLM Ranking. arXiv:2508.09539 [cs.IR] https://arxiv.org/abs/ 2508.09539

work page arXiv 2025
[11]

Thibault Formal, Benjamin Piwowarski, and Stéphane Clinchant. 2021. SPLADE: Sparse Lexical and Expansion Model for First Stage Ranking. InProceed- ings of the 44th International ACM SIGIR Conference on Research and Develop- ment in Information Retrieval(<conf-loc>, <city>Virtual Event</city>, <coun- try>Canada</country>, </conf-loc>)(SIGIR ’21). Associati...

work page doi:10.1145/3404835.3463098 2021
[12]

Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, et al . 2025. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning.arXiv preprint arXiv:2501.12948(2025)

work page internal anchor Pith review arXiv 2025
[13]

Mathew Jacob, Erik Lindgren, Matei Zaharia, Michael Carbin, Omar Khattab, and Andrew Drozdov. 2024. Drowning in documents: consequences of scaling reranker inference.arXiv preprint arXiv:2411.11767(2024)

work page arXiv 2024
[14]

Jardine and Cornelis Joost van Rijsbergen

N. Jardine and Cornelis Joost van Rijsbergen. 1971. The use of hierarchic clustering in information retrieval.Inf. Storage Retr.7, 5 (1971), 217–240. doi:10.1016/0020-0271(71)90051-9

work page doi:10.1016/0020-0271(71)90051-9 1971
[15]

Vladimir Karpukhin, Barlas Oguz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih. 2020. Dense Passage Retrieval for Open- Domain Question Answering. InProceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), Bonnie Webber, Trevor Cohn, Yulan He, and Yang Liu (Eds.). Association for Com...

work page doi:10.18653/v1/2020.emnlp-main.550 2020
[16]

Julian Killingback and Hamed Zamani. 2025. Benchmarking Information Retrieval Models on Complex Retrieval Tasks.arXiv preprint arXiv:2509.07253(2025)

work page arXiv 2025
[17]

Jongho Kim, Jaeyoung Kim, Seung-won Hwang, Jihyuk Kim, Yu Jin Kim, and Moontae Lee. 2026. Adaptive Retrieval for Reasoning-Intensive Retrieval.arXiv preprint arXiv:2601.04618(2026)

work page internal anchor Pith review Pith/arXiv arXiv 2026
[18]

Hrishikesh Kulkarni, Nazli Goharian, Ophir Frieder, and Sean MacAvaney. 2024. LexBoost: Improving Lexical Document Retrieval with Nearest Neighbors. In Proceedings of the ACM Symposium on Document Engineering 2024(San Jose, CA, USA)(DocEng ’24). Association for Computing Machinery, New York, NY, USA, Article 16, 10 pages. doi:10.1145/3685650.3685658

work page doi:10.1145/3685650.3685658 2024
[19]

Hrishikesh Kulkarni, Sean MacAvaney, Nazli Goharian, and Ophir Frieder. 2023. Lexically-Accelerated Dense Retrieval. InProceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2023, Taipei, Taiwan, July 23-27, 2023, Hsin-Hsi Chen, Wei-Jou (Edward) Duh, Hen-Hsen Huang, Makoto P. Kato, Josiane Mo...

work page doi:10.1145/3539618.3591715 2023
[20]

Junwei Lan, Jianlyu Chen, Zheng Liu, Chaofan Li, Siqi Bao, and Defu Lian
[21]

InThe Fourteenth International Conference on Learning Representations

Retro*: Optimizing LLMs for Reasoning-Intensive Document Retrieval. InThe Fourteenth International Conference on Learning Representations. https: //openreview.net/forum?id=0WGl8PNMSA
[22]

Sangam Lee, Ryang Heo, SeongKu Kang, and Dongha Lee. 2025. Imagine All The Relevance: Scenario-Profiled Indexing with Knowledge Expansion for Dense Retrieval.arXiv preprint arXiv:2503.23033(2025)

work page arXiv 2025
[23]

Sheng-Chieh Lin, Jheng-Hong Yang, and Jimmy Lin. 2021. In-Batch Nega- tives for Knowledge Distillation with Tightly-Coupled Teachers for Dense Re- trieval. InProceedings of the 6th Workshop on Representation Learning for NLP, RepL4NLP@ACL-IJCNLP 2021, Online, August 6, 2021, Anna Rogers, Iacer Calixto, Ivan Vulic, Naomi Saphra, Nora Kassner, Oana-Maria Ca...

work page doi:10.18653/v1/2021.repl4nlp-1.17 2021
[24]

Xueguang Ma, Liang Wang, Nan Yang, Furu Wei, and Jimmy Lin. 2024. Fine- tuning llama for multi-stage text retrieval. InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. 2421–2425

2024
[25]

Sean MacAvaney, Franco Maria Nardini, Raffaele Perego, Nicola Tonellotto, Nazli Goharian, and Ophir Frieder. 2020. Expansion via Prediction of Importance with Contextualization. InProceedings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval, SIGIR 2020, Virtual Event, China, July 25-30, 2020, Jimmy X. Hua...

work page arXiv 2020
[26]

Sean MacAvaney, Nicola Tonellotto, and Craig Macdonald. 2022. Adaptive Re- Ranking with a Corpus Graph. InProceedings of the 31st ACM International Conference on Information & Knowledge Management, Atlanta, GA, USA, October 17-21, 2022, Mohammad Al Hasan and Li Xiong (Eds.). ACM, 1491–1500. doi:10. 1145/3511808.3557231

work page arXiv 2022
[27]

Mackenzie, and Torsten Suel

Antonio Mallia, Michal Siedlaczek, Joel M. Mackenzie, and Torsten Suel. 2019. PISA: Performant Indexes and Search for Academia. InProceedings of the Open- Source IR Replicability Challenge co-located with 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval, OSIRRC@SIGIR 2019, Paris, France, July 25, 2019 (CEUR Work...

2019
[28]

Niklas Muennighoff, Hongjin SU, Liang Wang, Nan Yang, Furu Wei, Tao Yu, Aman- preet Singh, and Douwe Kiela. 2025. Generative Representational Instruction Tuning. InThe Thirteenth International Conference on Learning Representations. https://openreview.net/forum?id=BC4lIvfSzv

2025
[29]

Rodrigo Frassetto Nogueira, Zhiying Jiang, Ronak Pradeep, and Jimmy Lin. 2020. Document Ranking with a Pretrained Sequence-to-Sequence Model. InFindings of the Association for Computational Linguistics: EMNLP 2020, Online Event, 16-20 November 2020 (Findings of ACL, Vol. EMNLP 2020), Trevor Cohn, Yulan He, and Yang Liu (Eds.). Association for Computationa...

2020
[30]

Ronak Pradeep, Sahel Sharifymoghaddam, and Jimmy Lin. 2023. RankZephyr: Effective and Robust Zero-Shot Listwise Reranking is a Breeze!arXiv preprint arXiv:2312.02724(2023)

work page arXiv 2023
[31]

Mandeep Rathee, Sean MacAvaney, and Avishek Anand. 2025. Guiding Retrieval Using LLM-Based Listwise Rankers. InAdvances in Information Retrieval, Claudia Hauff, Craig Macdonald, Dietmar Jannach, Gabriella Kazai, Franco Maria Nardini, Fabio Pinelli, Fabrizio Silvestri, and Nicola Tonellotto (Eds.). Springer Nature Switzerland, Cham, 230–246

2025
[32]

Mandeep Rathee, Sean MacAvaney, and Avishek Anand. 2025. Quam: Adaptive Retrieval through Query Affinity Modelling. InProceedings of the Eighteenth ACM International Conference on Web Search and Data Mining(Hannover, Germany) (WSDM ’25). Association for Computing Machinery, New York, NY, USA, 954–962. doi:10.1145/3701551.3703584

work page doi:10.1145/3701551.3703584 2025
[33]

Mandeep Rathee, Venktesh V, Sean MacAvaney, and Avishek Anand. 2025. Breaking the Lens of the Telescope: Online Relevance Estimation over Large Reproducing Adaptive Reranking for Reasoning-Intensive IR Conference’17, July 2017, Washington, DC, USA Retrieval Sets. InProceedings of the 48th International ACM SIGIR Conference on Research and Development in I...

work page doi:10.1145/3726302.3729910 2025
[34]

Stephen Robertson. 2008. On the history of evaluation in IR.J. Inf. Sci.34, 4 (2008), 439–456. doi:10.1177/0165551507086989

work page doi:10.1177/0165551507086989 2008
[35]

Stephen Robertson and Hugo Zaragoza. 2009. The Probabilistic Relevance Framework: BM25 and Beyond.Found. Trends Inf. Retr.3, 4 (apr 2009), 333–389. doi:10.1561/1500000019

work page doi:10.1561/1500000019 2009
[36]

Harrisen Scells, Shengyao Zhuang, and Guido Zuccon. 2022. Reduce, Reuse, Recycle: Green Information Retrieval Research. InProceedings of the 45th In- ternational ACM SIGIR Conference on Research and Development in Information Retrieval(Madrid, Spain)(SIGIR ’22). Association for Computing Machinery, New York, NY, USA, 2825–2837. doi:10.1145/3477495.3531766

work page doi:10.1145/3477495.3531766 2022
[37]

Nilanjan Sinhababu, Andrew Parry, Debasis Ganguly, Debasis Samanta, and Pabitra Mitra. 2024. Few-shot Prompting for Pairwise Ranking: An Effective Non-Parametric Retrieval Model. InFindings of the Association for Computational Linguistics: EMNLP 2024, Yaser Al-Onaizan, Mohit Bansal, and Yun-Nung Chen (Eds.). Association for Computational Linguistics, Miam...

work page doi:10.18653/v1/2024.findings-emnlp.720 2024
[38]

Hongjin SU, Howard Yen, Mengzhou Xia, Weijia Shi, Niklas Muennighoff, Han yu Wang, Liu Haisu, Quan Shi, Zachary S Siegel, Michael Tang, Ruoxi Sun, Jinsung Yoon, Sercan O Arik, Danqi Chen, and Tao Yu. 2025. BRIGHT: A Realistic and Challenging Benchmark for Reasoning-Intensive Retrieval. InThe Thirteenth International Conference on Learning Representations....

2025
[39]

Weiwei Sun, Lingyong Yan, Xinyu Ma, Shuaiqiang Wang, Pengjie Ren, Zhumin Chen, Dawei Yin, and Zhaochun Ren. 2023. Is ChatGPT good at search? investigat- ing large language models as re-ranking agents.arXiv preprint arXiv:2304.09542 (2023)

work page arXiv 2023
[40]

Nandan Thakur, Jimmy Lin, Sam Havens, Michael Carbin, Omar Khattab, and Andrew Drozdov. 2025. FreshStack: Building Realistic Benchmarks for Evaluating Retrieval on Technical Documents. InThe Thirty-ninth Annual Conference on Neural Information Processing Systems Datasets and Benchmarks Track. https: //openreview.net/forum?id=54TTgXlS2U

2025
[41]

Venktesh V, Mandeep Rathee, and Avishek Anand. 2025. SUNAR: Semantic Uncertainty based Neighborhood Aware Retrieval for Complex QA. InProceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers), Luis Chiruzzo, Alan Ritter, and Lu Wang (Eds.)...

work page doi:10.18653/v1/2025.naacl- 2025
[42]

Lidan Wang, Jimmy Lin, and Donald Metzler. 2011. A cascade ranking model for efficient ranked retrieval. InProceedings of the 34th international ACM SIGIR conference on Research and development in Information Retrieval. 105–114

2011
[43]

Shuai Wang, Shengyao Zhuang, and Guido Zuccon. 2021. Bert-based dense retrievers require interpolation with bm25 for effective passage retrieval. In Proceedings of the 2021 ACM SIGIR international conference on theory of information retrieval. 317–324

2021
[44]

Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, brian ichter, Fei Xia, Ed Chi, Quoc V Le, and Denny Zhou. 2022. Chain-of-Thought Prompting Elicits Reasoning in Large Language Models. InAdvances in Neural Information Processing Systems, S. Koyejo, S. Mohamed, A. Agar- wal, D. Belgrave, K. Cho, and A. Oh (Eds.), Vol. 35. Curran Associates, Inc., 248...

2022
[45]

Orion Weller, Benjamin Chang, Sean MacAvaney, Kyle Lo, Arman Cohan, Ben- jamin Van Durme, Dawn Lawrie, and Luca Soldaini. 2025. FollowIR: Evalu- ating and Teaching Information Retrieval Models to Follow Instructions. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Languag...

work page doi:10.18653/v1/2025.naacl-long.597 2025
[46]

Lawrie, Ashwin Paranjape, Yuhao Zhang, and Jack Hessel

Orion Weller, Benjamin Van Durme, Dawn J. Lawrie, Ashwin Paranjape, Yuhao Zhang, and Jack Hessel. 2025. Promptriever: Instruction-Trained Retrievers Can Be Prompted Like Language Models. InThe Thirteenth International Conference on Learning Representations, ICLR 2025, Singapore, April 24-28, 2025. OpenReview.net. https://openreview.net/forum?id=odvSjn416y

2025
[47]

Orion Weller, Kathryn Ricci, Eugene Yang, Andrew Yates, Dawn Lawrie, and Ben- jamin Van Durme. 2025. Rank1: Test-time compute for reranking in information retrieval.arXiv preprint arXiv:2502.18418(2025)

work page arXiv 2025
[48]

Soyoung Yoon, Jongho Kim, Daeyong Kwon, Avishek Anand, and Seung-won Hwang. 2025. On Listwise Reranking for Corpus Feedback.arXiv preprint arXiv:2510.00887(2025)

work page arXiv 2025
[49]

Le Zhang, Bo Wang, Xipeng Qiu, Siva Reddy, and Aishwarya Agrawal. 2025. REARANK: Reasoning Re-ranking Agent via Reinforcement Learning. InProceed- ings of the 2025 Conference on Empirical Methods in Natural Language Processing, Christos Christodoulopoulos, Tanmoy Chakraborty, Carolyn Rose, and Violet Peng (Eds.). Association for Computational Linguistics,...

2025
[50]

doi:10.18653/v1/2025.emnlp-main.125

work page doi:10.18653/v1/2025.emnlp-main.125 2025
[51]

Yanzhao Zhang, Mingxin Li, Dingkun Long, Xin Zhang, Huan Lin, Baosong Yang, Pengjun Xie, An Yang, Dayiheng Liu, Junyang Lin, Fei Huang, and Jingren Zhou
[52]

Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models.arXiv preprint arXiv:2506.05176(2025)

work page internal anchor Pith review arXiv 2025
[53]

Shengyao Zhuang, Xueguang Ma, Bevan Koopman, Jimmy Lin, and Guido Zuccon
[54]

Rank-r1: Enhancing reasoning in llm-based document rerankers via reinforcement learning, 2025

Rank-R1: Enhancing Reasoning in LLM-based Document Rerankers via Reinforcement Learning. arXiv:2503.06034 [cs.IR] https://arxiv.org/abs/2503. 06034

work page arXiv