Millions of $\text{GeAR}$-s: Extending GraphRAG to Millions of Documents

Chenxin Diao; Jeff Z. Pan; Pascual Merita; Pavlos Vougiouklis; Zhili Shen

REVIEW 3 major objections 5 minor 14 references

Millions of $\text{GeAR}$-s: Extending GraphRAG to Millions of Documents

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper shows that graph-based RAG can be extended to millions of documents by replacing offline LLM triple extraction with an online, sparse-retrieval alignment to Wikidata triples, reporting correctness 0.875714 and faithfulness…

desk verdict A candid challenge-paper that proposes a neat way to skip offline triple extraction, but never shows the graph part actually helps; the case study is informative enough to warrant engagement. read the letter →

arxiv 2507.17399 v1 pith:WE3FYVOQ submitted 2025-07-23 cs.CL cs.AIcs.IR

Zhili Shen , Chenxin Diao , Pascual Merita , Pavlos Vougiouklis , Jeff Z. Pan This is my paper

classification cs.CLcs.AIcs.IR

keywords GraphRAGretrieval-augmentedgenerationmulti-hopquestionansweringWikidataknowledgealignmentsparseretrievalLivelargelanguagemodels

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Graph-based retrieval-augmented generation usually pays for multi-hop reasoning by running an LLM over every passage to extract triples, a cost that becomes prohibitive on millions of documents. This paper tries to show that the triple extraction step can be skipped entirely: the triples are read only from the few passages retrieved for a question, then linked on the fly to the nearest Wikidata triples using sparse retrieval, and those linked triples drive graph expansion to pull in more distant evidence. The authors report that this online approach, applied to the SIGIR 2025 LiveRAG corpus, achieves a correctness of 0.875714 and a faithfulness of 0.529335. If the approach holds up, it would remove the main cost barrier to applying GraphRAG at web scale.

What carries the argument

The load-bearing mechanism is the online triple-alignment step, Equation 3: each triple read from the retrieved chunks ($T'_{q(n)}$) is linked to the most similar triple in Wikidata by sparse retrieval ($h^1_{sparse}$). Linked triples are expanded into reasoning chains via GeAR's diverse triple beam search, and each triple in the beams is then mapped back to corpus chunks using another sparse-retrieval step. The candidate passages from graph expansion are fused with the baseline retrieval by Reciprocal Rank Fusion, Equation 4, and a filtering prompt removes irrelevant passages before the reader answers.

What would settle it

A reader could test the claim by running the full system on multi-hop questions with graph expansion from Wikidata enabled versus disabled; if the expanded retrieval does not improve answer correctness over the baseline hybrid retrieval alone, the central claim that online-linked Wikidata triples carry the multi-hop signal is falsified. A cheaper check is to inspect the linked triples for a sample of questions and count how often the topic shifts as in Table 2.

Watch

Extended reading notes

Core claim

The central claim is that proximal triples—atomic facts read from a small set of retrieved chunks—can be linked, through a simple sparse-retrieval step, to triples in an external knowledge graph such as Wikidata, and that these linked triples can then drive graph expansion to retrieve passages that support multi-hop reasoning. This makes triple extraction online and corpus-agnostic: the knowledge graph is not built from the corpus but borrowed. The paper reports that the resulting system, a modified GeAR, achieves correctness 0.875714 and faithfulness 0.529335 on the LiveRAG benchmark. The authors also present a case study showing that the sparse linking often misaligns topics—for instance, 'pacific geoducks' links to 'Pacific oyster'—which they take as evidence that the proxy assumption behind their approach is fragile and that better semantic models operating in a shared space for text and graph data are needed.

Load-bearing premise

The load-bearing premise is that a sparse-retrieval match between a triple read from a retrieved chunk and a triple in Wikidata is semantically faithful enough that expanding the graph from the linked triples brings in useful evidence rather than noise; the paper's own case study shows two examples where the match shifts topic.

Editorial extensions

If this is right

GraphRAG becomes applicable to corpora of millions of passages without the need to run LLM-based triple extraction over every passage, removing the main scaling bottleneck.
The reported scores (correctness 0.875714, faithfulness 0.529335) define a concrete performance reference for graph-enhanced retrieval on the LiveRAG benchmark.
Skipping graph expansion for the first retrieval step and applying it only to later, multi-hop iterations reduces compute while keeping graph benefits where they matter.
The demonstrated misalignment implies that improving the triple-linking model, not the graph expansion itself, is the next lever for better performance.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The same online alignment could be applied to any external knowledge graph and any dense/sparse passage index, making the approach a generic retrofit for existing GraphRAG pipelines that currently rely on offline extraction.
A natural follow-up would be replacing the sparse linker with a cross-encoder or an embedding model trained on (text, triple) pairs; if correctness rises substantially, the sparse linker, not the idea of borrowing Wikidata triples, is the bottleneck.
If the proxy assumption fails in general, a middle path would build a small task-specific knowledge graph from only the retrieved passages using an LLM at query time, preserving the scaling advantage while avoiding topic drift.
The gap between correctness and faithfulness suggests some final answers are right but not supported by the retrieved evidence; tracing whether the unsupported answers come from baseline or graph-expanded passages would isolate where the alignment noise hurts.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this is an honest challenge-paper about adapting GeAR to millions of documents by replacing offline LLM triple extraction with online sparse alignment to Wikidata. The new piece is Eq. (3): proximal triples read from retrieved chunks are linked to Wikidata triples via sparse retrieval, and those linked triples drive graph expansion. That is a legitimate cost-saving mechanism, and the paper deserves credit for reporting a case study (Table 2) that shows the linking frequently drifts off-topic, and for explicitly saying this challenges GeAR's key proxy assumption.

The soft spot is the one you'd expect: the 'simple yet effective' claim is not backed by evidence. The paper reports only two end-to-end scores (correctness 0.876, faithfulness 0.529) with no baseline, no ablation, no error bars, and no code. Since Wikidata triples are not used in the first iteration, graph expansion only affects second-iteration multi-hop questions. The two case-study examples suggest link precision is low enough that expanded passages may be noise; the reported scores could be entirely due to the hybrid retriever plus Falcon reader. So the stress-test note holds: the paper never demonstrates the graph component adds retrieval value, and its own evidence suggests it may subtract value. That is load-bearing for the effectiveness claim, though not fatal for a challenge report.

Minor: hyperparameters (max_steps_n, aliases_per_entity) were tuned on a DataMorgana dev set, and there is no breakdown of which question types benefit. I'd want to see that in any revision.

Who this is for: people working on GraphRAG scaling, RAG at web scale, and challenge participants. It deserves a serious referee because the mechanism is real and the writing is candid; the referee should ask for an ablation of Eqs. (3)-(4), a graph-free baseline, and confidence intervals. My own verdict is skeptical on the headline claim, but this is a useful system description and a good discussion piece. Send it to review with a request for revision.

Referee Report

3 major / 5 minor

Summary. This paper describes the authors' submission of a GraphRAG system, 'Graph-Enhanced RAG', to the SIGIR 2025 LiveRAG Challenge, which uses the FineWeb-10BT corpus containing millions of passages. The authors adapt their prior GeAR system by replacing the offline LLM-based triple extraction step with an online pseudo-alignment procedure: proximal triples read from baseline-retrieved chunks are linked to Wikidata triples via sparse retrieval (Eq. 3), and those triples are used to retrieve additional passages at later reasoning steps (Eq. 4). The system also includes query rewriting, passage filtering, and a final reader. The paper reports preliminary challenge scores of correctness 0.875714 and faithfulness 0.529335, and a case study showing two instances where the Wikidata linking step shifts topic. The central claim is that this online alignment approach is 'simple yet effective' and that it identifies limitations of current graph-text alignment methods.

Significance. If the effectiveness claim were established, the paper would make a useful contribution: it shows a way to run graph-based RAG at the scale of millions of documents without the prohibitive cost of offline LLM-based triple extraction, and it makes concrete falsifiable predictions about the need for shared semantic-space models for graph and text. The paper is honest about its limitations, and the two worked failure cases in Table 2 are valuable negative evidence for the research community. However, the central claim is not supported by the reported experiments: there is no baseline, no ablation of the new alignment step, no error bars, and no per-category breakdown. The case study itself suggests that the linking step can introduce off-topic passages, so the reported scores may not reflect any benefit from the graph component. The contribution is therefore currently an interesting system description with an unsupported effectiveness claim.

major comments (3)

[Section 5 and Section 6] The paper's only quantitative evidence is the pair of challenge scores reported in Section 1 (correctness 0.875714, faithfulness 0.529335), with no comparison to a graph-free baseline, no ablation of Eqs. (3)-(4), and no error bars or multiple runs. Since the Wikidata-based graph expansion is the only new component and Section 5 states that it is skipped during the first iteration, the reader cannot determine whether graph expansion improves, degrades, or is simply unused in the final scores. The 'simple yet effective' claim in Sections 1 and 7 is therefore not established; the authors should report scores for the hybrid retriever alone, for the full system, and ideally for a control condition in which the graph-expanded passages are replaced by randomly retrieved passages.
[Section 6 and Eq. (3)] The case study in Table 2 directly undermines the load-bearing assumption that proximal FineWeb triples can be reliably linked to Wikidata triples via sparse retrieval. In both examples, the linked Wikidata triples shift topic (from 'pacific geoducks' to 'pacific oyster', and from a hot tub heater to geography/biology items), and the manuscript itself states that this 'challenges a key assumption of the original GeAR system'. Because Eq. (4) uses passages retrieved from these linked triples, such misalignments can inject noise into the retrieval context. Without quantifying the precision of the Eq. (3) linking step on a sample and without showing that passages introduced through Eq. (4) improve or even affect the final answer, the reported scores may be entirely attributable to the hybrid retriever and the Falcon reader rather than to the graph component.
[Section 5.1] The evaluation is conducted on a development set constructed with DataMorgana, and the paper states that hyperparameters were chosen based on this set, but it does not report results on a held-out test set or break down results by question category. This matters because the DataMorgana taxonomy in Table 1 assigns only 30% of questions to the multi-hop categories (path-following and path-finding), and Section 5 explicitly disables graph expansion for the first iteration. Thus the aggregate scores may be dominated by single-hop questions that do not exercise the proposed graph alignment at all. The authors should report correctness and faithfulness separately for multi-hop and single-hop subsets, and should specify which hyperparameter choices were tuned on the development set.

minor comments (5)

[Section 1 vs. Section 3] The abstract and introduction refer to 'Falcon-3B-Instruct' as the knowledge synchronizer, while Section 3 and the prompts in Appendix A refer to 'Falcon3-10B-Instruct' as the reader; please clarify which model is used for reading, rewriting, and answering, since the two names denote different model sizes.
[Eq. (3)] The notation in Eq. (3) is inconsistent: the left-hand side is written Tq( j) with index j, while the text and surrounding equations use n for the current step; also, the function h1_sparse was previously defined for passages, and its application to triples should be specified more precisely.
[Table 2] The table refers to green and red keywords to indicate topics, but this is not accessible in monochrome print or for color-blind readers; please use explicit labels or markers in addition to color.
[Appendix A.2] There are several typos in the prompt listing, for example 'which we be think are relevant' should be 'which we think are relevant'.
[Section 6] The phrase 'topic shifts from toiletry machinery' is misleading, since a hot tub heater is not toiletry; consider rephrasing to 'domestic hot-water equipment' or similar.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper reports an externally evaluated empirical system, and its self-citations are implementation references rather than evidence that reduces the central claim to its own inputs.

full rationale

The paper's derivation chain is an empirical pipeline, not a deductive one: baseline hybrid retrieval (Eq. 1), LLM-based proximal triple extraction (Eq. 2), sparse linking of those triples to Wikidata (Eq. 3), graph expansion and sparse mapping back to passages (Eq. 4), query rewriting, filtering, and answer generation (Eqs. 5-7). No parameter is fitted to the reported correctness and faithfulness scores and then renamed a prediction; those scores come from the SIGIR 2025 LiveRAG Challenge's external automatic evaluation, which is outside the paper's own fitted values. The central 'simple yet effective' claim is an empirical engineering assertion, and the paper's own Section 6 case study explicitly documents failures of the Eq. 3 linking step, so the paper does not hide the assumption that proximal triples are good proxies for Wikidata triples. The self-citation to the authors' prior GeAR paper [7] is used for implementation details such as 'diverse triple beam search (see Section 4.2 in [7])', but it is not invoked as the justification for the effectiveness claim, and GeAR is an externally published system. There is no uniqueness theorem, no ansatz smuggled in via citation, and no renaming of a known result. The strongest criticism available from the text is that the contribution lacks an ablation isolating Eq. 3's effect, which is an evaluation gap rather than a circularity. Accordingly, no circular step can be exhibited with specific equations or quoted text, and the appropriate finding is no significant circularity.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The paper's central claim depends on the reliability of automatic fact extraction and sparse linking to Wikidata, plus the challenge evaluation as a proxy for quality. The method has no invented entities. The main engineered choices are the step limit, the decision to skip graph expansion in the first iteration, and the single-alias policy, all disclosed as development-set or cost-driven decisions.

free parameters (3)

max_steps_n = 2
Maximum number of agentic iterations; set after experiments because graph expansion gave limited benefit for simpler DataMorgana questions (Section 5).
graph_expansion_in_first_iteration = disabled
Chosen in Section 5 because graph expansion for simpler questions was limited; steps beyond the first use the full Wikidata pipeline.
aliases_per_entity = 1
Each Wikidata entity gets exactly one alias triple; more aliases were omitted to stay within compute credits (footnote 2).

assumptions (4)

domain assumption Proximal triples extracted from retrieved chunks are reliable proxies for triples in the external knowledge graph
Stated as the original GeAR assumption in Section 6, and explicitly challenged by the paper's own case study.
domain assumption Sparse lexical retrieval can align natural-language facts to Wikidata triples
Used in Eq. 3 for linking T' to T; the Discussion shows topic drift, e.g., geoducks to Pacific oysters.
domain assumption Wikidata covers the entity and relation space needed for diverse web questions
The full Wikidata dump is used as the external triple index (Section 5) without coverage analysis.
domain assumption Falcon3-10B-Instruct performs read, rewrite, filter, and answer steps correctly enough with the provided prompts
Used as the reader and answerer (Sections 3 and 4); no accuracy or steering validation is reported.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Millions of $\text{GeAR}$-s: Extending GraphRAG to Millions of Documents." pith.science (2026). https://pith.science/paper/WE3FYVOQ

@misc{pith2026250717399,
  author       = {Pith},
  title        = {Pith review of: Millions of $\textGeAR$-s: Extending GraphRAG to Millions of Documents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WE3FYVOQ}},
  note         = {Machine review of arXiv:2507.17399}
}

abstract

Recent studies have explored graph-based approaches to retrieval-augmented generation, leveraging structured or semi-structured information -- such as entities and their relations extracted from documents -- to enhance retrieval. However, these methods are typically designed to address specific tasks, such as multi-hop question answering and query-focused summarisation, and therefore, there is limited evidence of their general applicability across broader datasets. In this paper, we aim to adapt a state-of-the-art graph-based RAG solution: $\text{GeAR}$ and explore its performance and limitations on the SIGIR 2025 LiveRAG Challenge.

Figures

Figures reproduced from arXiv: 2507.17399 by the authors.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

14 extracted references · 11 canonical work pages

[1]

Cormack, Charles L A Clarke, and Stefan Buettcher

Gordon V. Cormack, Charles L A Clarke, and Stefan Buettcher. 2009. Recip- rocal rank fusion outperforms condorcet and individual rank learning meth- ods. In Proceedings of the 32nd International ACM SIGIR Conference on Research and Development in Information Retrieval (Boston, MA, USA) (SIGIR ’09) . As- sociation for Computing Machinery, New York, NY, USA...

arXiv 2009
[2]

Jinyuan Fang, Zaiqiao Meng, and Craig MacDonald. 2024. TRACE the Evidence: Constructing Knowledge-Grounded Reasoning Chains for Retrieval-Augmented Generation. In Findings of the Association for Computational Linguistics: EMNLP 2024, Yaser Al-Onaizan, Mohit Bansal, and Yun-Nung Chen (Eds.). Association for Computational Linguistics, Miami, Florida, USA, 8...

work page 2024
[3]

Simone Filice, Guy Horowitz, David Carmel, Zohar Karnin, Liane Lewin-Eytan, and Yoelle Maarek. 2025. Generating Diverse Q&A Benchmarks for RAG Evaluation with DataMorgana. arXiv:2501.12789 [cs.CL] https://arxiv.org/abs/2501.12789

arXiv 2025
[4]

Bernal Jimenez Gutierrez, Yiheng Shu, Yu Gu, Michihiro Yasunaga, and Yu Su. 2024. HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language Models. In The Thirty-eighth Annual Conference on Neural Information Processing Systems. https://openreview.net/forum?id=hkujvAPVsg

work page 2024
[5]

Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. 2020. Retrieval-augmented generation for Millions of GeAR-s : Extending GraphRAG to Millions of Documents SIGIR 2025 LiveRAG, July 2025, Padua, Italy knowledge-intensi...

work page 2020
[6]

Shilong Li, Yancheng He, Hangyu Guo, Xingyuan Bu, Ge Bai, Jie Liu, Jiaheng Liu, Xingwei Qu, Yangguang Li, Wanli Ouyang, Wenbo Su, and Bo Zheng. 2024. GraphReader: Building Graph-based Agent to Enhance Long-Context Abilities of Large Language Models. In Findings of the Association for Computational Lin- guistics: EMNLP 2024, Yaser Al-Onaizan, Mohit Bansal,...

work page 2024
[7]

Graph-Enhanced RAG

Zhili Shen, Chenxin Diao, Pavlos Vougiouklis, Pascual Merita, Shriram Pira- manayagam, Damien Graux, Dandan Tu, Zeren Jiang, Ruofei Lai, Yang Ren, and Jeff Z. Pan. 2024. GeAR: Graph-enhanced Agent for Retrieval-augmented Genera- tion. arXiv:2412.18431 [cs.CL] https://arxiv.org/abs/2412.18431 SIGIR 2025 LiveRAG, July 2025, Padua, Italy Shen et al. A Prompt...

arXiv 2024
[8]

If any passages are irrelevant, please remove their identifier completely from results

> etc. If any passages are irrelevant, please remove their identifier completely from results. Return ‘None‘ if there are no relevant passages. We also give you a list of knowledge triples which be think are relevant to the query. Use them to help you rank the passages. SIGIR 2025 LiveRAG, July 2025, Padua, Italy Shen et al. ================== For example...

work page 2025

Show all 14 references

[9]

I Am a Fugitive from a Chain Gang

Wikipedia Title: Laughter in Hell Laughter in Hell is a 1933 American Pre-Code drama film directed by Edward L. Cahn and starring Pat O’Brien. The film’s title was typical of the sensationalistic titles of many Pre-Code films. Adapted from the 1932 novel of the same name buy J...

1933
[10]

The date of Theodred’s consecration unknown, but the date of his death was sometime between 995 and 997

Wikipedia Title: Theodred II (Bishop of Elmham) Theodred II was a medieval Bishop of Elmham. The date of Theodred’s consecration unknown, but the date of his death was sometime between 995 and 997
[11]

Cahn Edward L

Wikipedia Title: Edward L. Cahn Edward L. Cahn (February 12, 1899 – August 25, 1963) was an American film director

1963
[12]

What is God?

Wikipedia Title: Etan Boritzer Etan Boritzer (born 1950) is an American writer of children’s literature who is best known for his book “What is God?” first published in 1989. His best selling “What is?” illustrated children’s book series on character education and difficult su...

1950
[13]

Reranked Passages: [3] > [1] ================== The following are {num_docs} passages, each indicated by number identifier []

Wikipedia Title: Peter Levin Peter Levin is an American director of film, television and theatre. Reranked Passages: [3] > [1] ================== The following are {num_docs} passages, each indicated by number identifier []. Please rank them based on their relevance to query. ...
[14]

If any passages are irrelevant, please remove their identifier completely from results

> etc. If any passages are irrelevant, please remove their identifier completely from results. Return ‘None‘ if there are no relevant passages. Question: {query} Knowledge Triples: {triples} Passages: {retrieved_docs} Reranked Passages: Received 23 May 2025

2025

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

Cormack, Charles L A Clarke, and Stefan Buettcher

Gordon V. Cormack, Charles L A Clarke, and Stefan Buettcher. 2009. Recip- rocal rank fusion outperforms condorcet and individual rank learning meth- ods. In Proceedings of the 32nd International ACM SIGIR Conference on Research and Development in Information Retrieval (Boston, MA, USA) (SIGIR ’09) . As- sociation for Computing Machinery, New York, NY, USA...

arXiv 2009

[2] [2]

Jinyuan Fang, Zaiqiao Meng, and Craig MacDonald. 2024. TRACE the Evidence: Constructing Knowledge-Grounded Reasoning Chains for Retrieval-Augmented Generation. In Findings of the Association for Computational Linguistics: EMNLP 2024, Yaser Al-Onaizan, Mohit Bansal, and Yun-Nung Chen (Eds.). Association for Computational Linguistics, Miami, Florida, USA, 8...

work page 2024

[3] [3]

Simone Filice, Guy Horowitz, David Carmel, Zohar Karnin, Liane Lewin-Eytan, and Yoelle Maarek. 2025. Generating Diverse Q&A Benchmarks for RAG Evaluation with DataMorgana. arXiv:2501.12789 [cs.CL] https://arxiv.org/abs/2501.12789

arXiv 2025

[4] [4]

Bernal Jimenez Gutierrez, Yiheng Shu, Yu Gu, Michihiro Yasunaga, and Yu Su. 2024. HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language Models. In The Thirty-eighth Annual Conference on Neural Information Processing Systems. https://openreview.net/forum?id=hkujvAPVsg

work page 2024

[5] [5]

Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. 2020. Retrieval-augmented generation for Millions of GeAR-s : Extending GraphRAG to Millions of Documents SIGIR 2025 LiveRAG, July 2025, Padua, Italy knowledge-intensi...

work page 2020

[6] [6]

Shilong Li, Yancheng He, Hangyu Guo, Xingyuan Bu, Ge Bai, Jie Liu, Jiaheng Liu, Xingwei Qu, Yangguang Li, Wanli Ouyang, Wenbo Su, and Bo Zheng. 2024. GraphReader: Building Graph-based Agent to Enhance Long-Context Abilities of Large Language Models. In Findings of the Association for Computational Lin- guistics: EMNLP 2024, Yaser Al-Onaizan, Mohit Bansal,...

work page 2024

[7] [7]

Graph-Enhanced RAG

Zhili Shen, Chenxin Diao, Pavlos Vougiouklis, Pascual Merita, Shriram Pira- manayagam, Damien Graux, Dandan Tu, Zeren Jiang, Ruofei Lai, Yang Ren, and Jeff Z. Pan. 2024. GeAR: Graph-enhanced Agent for Retrieval-augmented Genera- tion. arXiv:2412.18431 [cs.CL] https://arxiv.org/abs/2412.18431 SIGIR 2025 LiveRAG, July 2025, Padua, Italy Shen et al. A Prompt...

arXiv 2024

[8] [8]

If any passages are irrelevant, please remove their identifier completely from results

> etc. If any passages are irrelevant, please remove their identifier completely from results. Return ‘None‘ if there are no relevant passages. We also give you a list of knowledge triples which be think are relevant to the query. Use them to help you rank the passages. SIGIR 2025 LiveRAG, July 2025, Padua, Italy Shen et al. ================== For example...

work page 2025