Pith. sign in

REVIEW 4 major objections 6 minor 34 references

Reasoning Paths as Signals: Augmenting Multi-hop Fact Verification through Structural Reasoning Progression

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper claims that explicitly modeling the evolving reasoning path as a sequence of evidence subgraphs improves multi-hop fact verification on FEVER and HoVer, with the largest gains on 4-hop claims.

desk verdict A solid incremental fact-verification paper whose progressive-subgraph idea is worth a look, but whose headline claim overreaches because it never compares against the authors' own MRR-FV baseline. read the letter →

arxiv 2506.07075 v1 pith:SWK6OKLS submitted 2025-06-08 cs.AI

Liwen Zheng , Chaozhuo Li , Haoran Jia , Xi Zhang This is my paper · ORCID
classification cs.AI
keywords multi-hopfactverificationevidenceretrievalreasoningpathstructure-enhancedqueryGraphFormersFEVERHoVerprogressivesubgraphs
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

The paper tries to establish that multi-hop fact verification improves when the reasoning path itself, not just the retrieved evidence, is treated as an explicit modeling object. Its framework, SR-MFV, builds a reasoning graph at each retrieval step to shape the next query, then turns the ordered evidence into a sequence of growing subgraphs for verification. A sympathetic reader should care because complex claims require evidence to be combined in a specific order and structure, and static or concatenation-based models miss that. The reported results on FEVER and HoVer support the claim: label accuracy and retrieval F1 are consistently higher than strong baselines, with the largest gap on 4-hop HoVer claims.

What carries the argument

The load-bearing object is the progressive subgraph sequence $\{G_1, G_2, \dots, G_n\}$, where each $G_k$ contains the claim and the first $k$ retrieved evidence sentences, preserving retrieval order as a proxy for logical inference order. Each subgraph's edges combine intra-sentence adjacency, inter-sentence coreference, and learned latent edges selected by thresholding pairwise similarities. The verification module reasons over each subgraph with GraphFormers, a Transformer variant that interleaves token-level self-attention with graph-structured aggregation, then fuses the per-subgraph representations with learned attention weights. On the retrieval side, an analogous reasoning graph over the claim and previously retrieved evidence undergoes graph attention message passing to produce the next-hop query, so that each retrieval step is grounded in the evolving evidence structure rather than in plain text concatenation.

What would settle it

Take a fixed set of retrieved evidence for a HoVer claim and reverse the order in which the verification module receives it. If SR-MFV's prediction and confidence stay essentially unchanged under reversal, then the progressive subgraph sequence is not carrying the claimed reasoning signal; if accuracy drops sharply, the order is load-bearing.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that the reasoning trajectory of a multi-hop verification system can be used as a modeling signal instead of being discarded after retrieval. The proposed SR-MFV builds a reasoning graph during retrieval to generate each next-hop query, then turns the ordered evidence list into a sequence of nested subgraphs, each containing the claim and the first $k$ evidence sentences. A GraphFormers encoder reasons over each subgraph, and an attention-weighted fusion of the subgraph states produces the final SUPPORTED, REFUTED, or NOT ENOUGH INFO prediction. On FEVER, this reaches 80.17 label accuracy and 77.62 FEVER score on the test set; on HoVer it reaches 79.05, 77.93, and 78.36 label accuracy on 2-, 3-, and 4-hop claims, the best in each column, with the largest advantage on 4-hop claims. The paper also reports retrieval F1 gains on both datasets, at the cost of slightly lower recall on FEVER due to a more conservative retrieval strategy.

Load-bearing premise

The load-bearing premise is that the order in which evidence sentences are retrieved is a faithful proxy for the order of logical inference, so that subgraphs built in retrieval order capture the true reasoning trajectory; if retrieval order is noisy, the progressive construction may not reflect genuine reasoning dynamics.

Editorial extensions

If this is right

  • If correct, preserving the original retrieval order through the verification stage is a useful inductive bias, so systems should stop destroying evidence order when building verification graphs.
  • Structure-enhanced query generation can make retrieval more precise: on FEVER the method raises Precision@5 and F1@5 over CO-GAT while accepting lower recall.
  • The optimal number of hops is dataset-dependent, roughly matching the annotated claim complexity on HoVer and peaking at two hops on FEVER, so deployed systems should tune retrieval depth rather than fix it.
  • GraphFormers-style interleaving of token attention and graph aggregation is a viable backbone for evidence-graph reasoning, not just for the original textual-graph task.

Reading between the lines

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

  • A natural extension outside the paper is to randomize or reverse retrieval order before verification; if the attention-weighted subgraph fusion still performs as well, the benefit comes from the graph encoder rather than from reasoning progression.
  • The subgraph attention weights from Equation (15) could be exposed as a per-hop explanation of which evidence moved the prediction, turning the verification module into an interpretability tool the paper does not build.
  • Because the retrieval graph and verification subgraphs share the same edge vocabulary, the approach might transfer to open-domain multi-hop question answering, where query decomposition and evidence ordering are similarly load-bearing.
  • A stress test of the latent edge threshold $\tau$ in Equation (7) would show whether performance is sensitive to the choice; if a broad range of $\tau$ works, the method is more likely to generalize.
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.

Referee Report

4 major / 6 minor

Summary. The paper proposes SR-MFV, a multi-hop fact verification framework with two modules: a structure-enhanced retrieval module that constructs a reasoning graph from the claim and previously retrieved evidence to guide the next-hop query, and a reasoning-path-guided verification module that incrementally builds subgraphs in retrieval order and processes them with GraphFormers to capture evolving inference trajectories. Experiments on FEVER and HoVer report improvements in label accuracy and retrieval F1 over the selected baselines, with ablations supporting the contribution of each component. The main claims are that explicitly modeling reasoning paths as graphs improves both retrieval precision and verification accuracy, particularly for 4-hop claims on HoVer.

Significance. If the reported results hold, the idea of representing retrieval order as a structural reasoning progression and feeding it into a graph-Transformer verification module is a plausible and potentially useful contribution to multi-hop fact verification. The paper is clearly written, the method is well specified, and the ablation study targets the key design choices. However, the significance is currently limited by the omission of the authors' own strong multi-hop system MRR-FV (ref. [35]) from the baseline comparison, by apparent inconsistencies between claimed gains and the numbers in Tables II and IV, and by the unvalidated assumption that retrieval order approximates logical inference order. These issues need to be addressed before the central claim of consistent superiority can be accepted.

major comments (4)
  1. [Section V-A/B, Tables I and II] The paper's headline claim that SR-MFV achieves the best overall performance and outperforms all compared baselines is not tested against the authors' own MRR-FV system (ref. [35]), which is a multi-hop retrieval-and-reasoning model specifically designed for complex fact verification and is cited in the Related Work. Since MRR-FV is arguably the most relevant recent baseline for HoVer, omitting it from Tables I and II and from the baseline list in Section V-A makes the claim 'outperforming all compared baselines' true only for a selected subset. Please add MRR-FV to the comparison, report per-hop HoVer results, and qualify the claim accordingly if MRR-FV's published scores are competitive or higher.
  2. [Section V-B and Section V-C] The numerical gains stated in the text do not match the corresponding tables. Section V-B reports consistent gains of 1.48%, 1.51%, and 3.09% on HoVer 2/3/4-hop label accuracy, but Table II gives differences between SR-MFV and the strongest listed baseline (SAGP) of 1.15, 1.15, and 2.35 percentage points. Similarly, Section V-C reports retrieval F1 gains of 0.49%, 0.73%, and 1.12% on 2/3/4-hop, whereas Table IV shows differences versus GMR of 0.4, 0.9, and 0.9 points and versus Baleen of 1.1, 0.6, and 1.1 points. Please correct the text or explicitly state which baseline and table these percentages refer to; as written, the claimed improvements are not reproducible from the reported results.
  3. [Section IV-B1, 'Subgraphs Construction'] The progressive subgraph construction is the core novelty of the verification module, but it rests on the stated assumption that the retrieval order is a proxy for the logical inference process. The paper provides no evidence for this assumption. If retrieval order is noisy, the sequence of subgraphs may not reflect a genuine reasoning trajectory, and any gains could come from the GraphFormer encoder rather than from the progressive structure. Please add a direct test: compare verification accuracy when evidence is fed in retrieval order, in gold-annotated order, and in a random order, and report whether the proposed progressive construction still helps. This would make the central mechanism falsifiable.
  4. [Section V-B and Section V-D] No uncertainty estimates or significance tests are reported. In Table V the ablation gaps are small (for example, 0.42 and 0.28 percentage points on the 2-hop subset for 'w/o Subgraphs Construction' and 'w/o GraphFormers'), and in Table II the claimed improvements over SAGP are 1.15 points on 2-hop and 3-hop. Without multiple runs, confidence intervals, or significance tests, it is unclear whether these differences are meaningful. Please report standard deviations across at least three seeds or provide a significance test for the main comparisons and ablations.
minor comments (6)
  1. [References] References [19] and [20] are identical (both are the Baleen paper), and references [26] and [27] duplicate reference [10] (SAGP). Please replace the duplicates with the intended distinct works or remove them.
  2. [Table IV] The table title uses 'HOVER' while the rest of the paper uses 'HoVer'; please make the notation consistent.
  3. [Section V-B] The text alternately uses 'SR-MFV' and 'MFV' (e.g., 'these results highlight that MFV's ability'); please use the full model name consistently.
  4. [Equation (7)] The threshold tau is described as 'tunable', but the paper does not report its value or sensitivity. Please state the value used and, ideally, add a sensitivity analysis for this hyperparameter.
  5. [Figure 3] The y-axis is labeled 'Accuracy' but the paper does not say whether the plotted values are dev or test accuracy; please specify the split in the caption or axis label.
  6. [Appendix/Notation] In Section IV-B2, the notation for the node representation h_g^(l) and its augmented version is not fully defined; in particular, Eq. (12) concatenates a vector with h_g^(l), but Eq. (9) defines z_g^(l) as a scalar [CLS] embedding. Please clarify the dimensional conventions.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the method's components are empirically evaluated and ablated, and the one definitional proxy (retrieval order as reasoning order) is explicitly acknowledged rather than presented as a derivation.

full rationale

The paper contains no derivation chain in which an output is defined in terms of the fitted input or renamed as a prediction. The retrieval module produces evidence via graph-attention query construction (Eqs. 1-5); the verification module consumes that evidence order to build subgraphs (Eq. 8) and reasons via GraphFormers (Eqs. 9-14), followed by attention fusion and an MLP classifier. This is a standard supervised pipeline evaluated on FEVER and HoVer against external baselines, with ablations (Table V) that vary the progressive construction, graph formation, and GraphFormers components. The only conceptual point that might look definitional is Section IV-B1's statement that retrieval order is used 'as a proxy for the logical inference process'; however, the paper explicitly labels this as a proxy, and the claimed benefit is tested empirically rather than derived from the definition. The citation of the authors' prior MRR-FV ([35]) is not load-bearing for any derivation, although its absence from the baseline tables is a comparison limitation that affects the strength of the 'best overall performance' headline, not the circularity of the method. No fitted parameter is renamed as a prediction, and no uniqueness theorem or self-citation chain is used to force a choice.

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

Most assumptions are standard domain assumptions of the ML approach, but the retrieval-order-as-reasoning assumption is a specific modeling choice central to the method and not independently justified.

free parameters (3)
  • tau = not stated
    Threshold for retaining learned semantic edges in E_sem; introduced in Equation (7). Value not reported.
  • number of retrieval hops = 2 for FEVER; 2/3/4 for HoVer 2/3/4-hop subsets
    Tuned per dataset, as shown in Figure 3.
  • number of GraphFormers layers L = not stated
    Depth of the graph reasoning stack; not specified in the paper.
assumptions (3)
  • domain assumption Retrieval order is a faithful proxy for logical inference order
    Section IV-B1 states subgraphs are built 'preserving the retrieval order as a proxy for the logical inference process'.
  • domain assumption Token-level graphs with adjacency, coreference, and learned edges capture the reasoning structure needed for fact verification
    Section IV-B1 defines edge types E_adj, E_coref, E_sem and assumes they are sufficient for multi-hop inference.
  • domain assumption GraphFormers provides superior reasoning over standard GNNs in this setting
    Section IV-B2 adopts GraphFormers based on prior work, assuming its structure-aware attention is beneficial.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Reasoning Paths as Signals: Augmenting Multi-hop Fact Verification through Structural Reasoning Progression." pith.science (2026). https://pith.science/paper/SWK6OKLS

@misc{pith2026250607075,
  author       = {Pith},
  title        = {Pith review of: Reasoning Paths as Signals: Augmenting Multi-hop Fact Verification through Structural Reasoning Progression},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SWK6OKLS}},
  note         = {Machine review of arXiv:2506.07075}
}
read the original abstract

The growing complexity of factual claims in real-world scenarios presents significant challenges for automated fact verification systems, particularly in accurately aggregating and reasoning over multi-hop evidence. Existing approaches often rely on static or shallow models that fail to capture the evolving structure of reasoning paths, leading to fragmented retrieval and limited interpretability. To address these issues, we propose a Structural Reasoning framework for Multi-hop Fact Verification that explicitly models reasoning paths as structured graphs throughout both evidence retrieval and claim verification stages. Our method comprises two key modules: a structure-enhanced retrieval mechanism that constructs reasoning graphs to guide evidence collection, and a reasoning-path-guided verification module that incrementally builds subgraphs to represent evolving inference trajectories. We further incorporate a structure-aware reasoning mechanism that captures long-range dependencies across multi-hop evidence chains, enabling more precise verification. Extensive experiments on the FEVER and HoVer datasets demonstrate that our approach consistently outperforms strong baselines, highlighting the effectiveness of reasoning-path modeling in enhancing retrieval precision and verification accuracy.

Figures

Figures reproduced from arXiv: 2506.07075 by the authors.

Figure 1
Figure 1. A complex claim requiring multi-hop reasoning over entities, events, [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Framework of the proposed SR-MFV model. where Enc(s) is the semantic embedding of candidate evidence s, and sim(·, ·) denotes a similarity function. By grounding the retrieval process in a semantically and structurally enriched reasoning graph, the model captures cross-evidence dependencies more effectively, enabling the construction of logically coherent reasoning paths. B. Reasoning-Path-Guided Claim Verification … view at source ↗
Figure 3
Figure 3. Hyperparameter sensitivity analysis(Accuracy vs. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 34 canonical work pages

  1. [35]

    MRR-FV: Unlocking Complex Fact Verification with Multi-hop Re- trieval and Reasoning,

    L. Zheng, C. Li, L. Zhang, H. Jia, S. Wang, Z. Liu, and X. Zhang, “MRR-FV: Unlocking Complex Fact Verification with Multi-hop Re- trieval and Reasoning,” in *Proc. AAAI Conf. Artif. Intell. (AAAI)*, 2025, vol. 39, no. 24, pp. 26066–26074

    work page 2025

  2. [1]

    FEVER: a large-scale dataset for fact extraction and verification,

    J. Thorne, A. Vlachos, C. Christodoulopoulos, and A. Mittal, “FEVER: a large-scale dataset for fact extraction and verification,” in Proc. Conf. North Amer. Chapter Assoc. Comput. Linguistics: Human Language Technol. (NAACL-HLT), 2018, pp. 809–819

    work page 2018

  3. [2]

    HoVer: a dataset for many-hop fact extraction and claim verification,

    Y . Jiang, S. Bordia, Z. Zhong, C. Dognin, M. K. Singh, and M. Bansal, “HoVer: a dataset for many-hop fact extraction and claim verification,” in Findings of the Assoc. Comput. Linguistics: EMNLP, 2020, pp. 3441–3460

    work page 2020

  4. [3]

    GEAR: graph-based evidence aggregating and reasoning for fact verification,

    J. Zhou, X. Han, C. Yang, Z. Liu, L. Wang, C. Li, and M. Sun, “GEAR: graph-based evidence aggregating and reasoning for fact verification,” in Proc. Annu. Meeting Assoc. Comput. Linguistics (ACL), vol. 1, 2019, pp. 892–901

    work page 2019

  5. [4]

    Fine-grained fact verification with kernel graph attention network,

    Z. Liu, C. Xiong, M. Sun, and Z. Liu, “Fine-grained fact verification with kernel graph attention network,” in Proc. Annu. Meeting Assoc. Comput. Linguistics (ACL), 2020, pp. 7342–7351

    work page 2020

  6. [5]

    Reasoning over semantic-level graph for fact checking,

    W. Zhong, J. Xu, D. Tang, Z. Xu, N. Duan, M. Zhou, J. Wang, and J. Yin, “Reasoning over semantic-level graph for fact checking,” in Proc. Annu. Meeting Assoc. Comput. Linguistics (ACL), 2020, pp. 6170–6180

    work page 2020

  7. [6]

    EvidenceNet: evidence fusion network for fact verification,

    Z. Chen, S. C. Hui, F. Zhuang, L. Liao, F. Li, M. Jia, and J. Li, “EvidenceNet: evidence fusion network for fact verification,” in Proc. Int. World Wide Web Conf. (WWW), 2022, pp. 2636–2645

    work page 2022

  8. [7]

    Hierarchical evidence set modeling for automated fact extraction and verification,

    S. Subramanian and K. Lee, “Hierarchical evidence set modeling for automated fact extraction and verification,” in Proc. Conf. Empiri- cal Methods Natural Language Process. (EMNLP), vol. 1, 2020, pp. 7798–7809

    work page 2020

Show all 34 references
  1. [8]

    Multi-evidence based fact verification via a confidential graph neural network,

    Y . Lan, Z. Liu, Y . Gu, X. Yi, X. Li, L. Yang, and G. Yu, “Multi-evidence based fact verification via a confidential graph neural network,” IEEE Trans. Big Data, vol. 11, 2025, pp. 426–437

    2025

  2. [9]

    DeBERTa: decoding-enhanced BERT with disentangled attention,

    P. He, X. Liu, J. Gao, and W. Chen, “DeBERTa: decoding-enhanced BERT with disentangled attention,” in Proc. Int. Conf. Learn. Represent. (ICLR), 2021

    2021

  3. [11]

    GERE: generative evidence retrieval for fact verification,

    J. Chen, R. Zhang, J. Guo, Y . Fan, and X. Cheng, “GERE: generative evidence retrieval for fact verification,” in Proc. Int. ACM SIGIR Conf. Res. Develop. Inf. Retrieval (SIGIR), 2022, pp. 2184–2189

    2022

  4. [12]

    Topic-aware evidence reasoning and stance-aware aggregation for fact verification,

    J. Si, D. Zhou, T. Li, X. Shi, and Y . He, “Topic-aware evidence reasoning and stance-aware aggregation for fact verification,” in Proc. Annu. Meeting Assoc. Comput. Linguistics / Int. Joint Conf. Natural Language Process. (ACL/IJCNLP), vol. 1, 2021, pp. 1612–1622

    2021

  5. [13]

    ProoFVer: natural logic theorem proving for fact verification,

    A. Krishna, S. Riedel, and A. Vlachos, “ProoFVer: natural logic theorem proving for fact verification,” Trans. Assoc. Comput. Linguistics, vol. 10, pp. 1013–1030, 2022

    2022

  6. [14]

    Combining fact extraction and verification with neural semantic matching networks,

    Y . Nie, H. Chen, and M. Bansal, “Combining fact extraction and verification with neural semantic matching networks,” in Proc. AAAI Conf. Artif. Intell. (AAAI), 2019, pp. 6859–6866

    2019

  7. [15]

    A multi-level attention model for evidence-based fact checking,

    C. Kruengkrai, J. Yamagishi, and X. Wang, “A multi-level attention model for evidence-based fact checking,” in Findings Assoc. Comput. Linguistics: ACL/IJCNLP, 2021, pp. 2447–2460

    2021

  8. [16]

    IMCI: integrate multi-view contextual information for fact extraction and verification,

    H. Wang, Y . Li, Z. Huang, and Y . Dou, “IMCI: integrate multi-view contextual information for fact extraction and verification,” in Proc. Int. Conf. Comput. Linguistics (COLING), 2022, pp. 1412–1421

    2022

  9. [17]

    UKP-Athene: multi-sentence textual entailment for claim verification,

    A. Hanselowski, H. Zhang, Z. Li, D. Sorokin, B. Schiller, C. Schulz, and I. Gurevych, “UKP-Athene: multi-sentence textual entailment for claim verification,” in Proc. Fact Extraction Verification Workshop (FEVER)@EMNLP, 2018, pp. 103–108

    2018

  10. [18]

    Claim-Dissector: an interpretable fact-checking system with joint re-ranking and veracity prediction,

    M. Fajcik, P. Motl ´ıcek, and P. Smrz, “Claim-Dissector: an interpretable fact-checking system with joint re-ranking and veracity prediction,” in Findings Assoc. Comput. Linguistics: ACL, 2023, pp. 10184–10205

    2023

  11. [20]

    Baleen: robust multi-hop reasoning at scale via condensed retrieval,

    O. Khattab, C. Potts, and M. A. Zaharia, “Baleen: robust multi-hop reasoning at scale via condensed retrieval,” in Adv. Neural Inf. Process. Syst. (NeurIPS), 2021, pp. 27670–27682

    2021

  12. [21]

    DeFormer: Decomposing pre-trained transformers for faster question answering,

    Q. Cao, H. Trivedi, A. Balasubramanian, and N. Balasubramanian, “DeFormer: Decomposing pre-trained transformers for faster question answering,” in Proc. Annu. Meeting Assoc. Comput. Linguistics (ACL), 2020, pp. 4487–4497

    2020

  13. [22]

    Exploring listwise evidence reasoning with T5 for fact verification,

    K. Jiang, R. Pradeep, and J. Lin, “Exploring listwise evidence reasoning with T5 for fact verification,” in Proc. Annu. Meeting Assoc. Comput. Linguistics / Int. Joint Conf. Natural Language Process. (ACL/IJCNLP), vol. 2, 2021, pp. 402–410

    2021

  14. [23]

    A survey on automated fact-checking,

    Z. Guo, M. S. Schlichtkrull, and A. Vlachos, “A survey on automated fact-checking,” Trans. Assoc. Comput. Linguistics, vol. 10, pp. 178–206, 2022

    2022

  15. [24]

    Autoregressive entity retrieval,

    N. De Cao, G. Izacard, S. Riedel, and F. Petroni, “Autoregressive entity retrieval,” in Proc. Int. Conf. Learn. Represent. (ICLR), 2021

    2021

  16. [25]

    Multi-hop fact checking of political claims,

    W. Ostrowski, A. Arora, P. Atanasova, and I. Augenstein, “Multi-hop fact checking of political claims,” in Proc. Int. Joint Conf. Artif. Intell. (IJCAI), 2021, pp. 3892–3898

    2021

  17. [26]

    Self-RAG: Learning to retrieve, generate, and critique through self-reflection,

    A. Asai, Z. Wu, Y . Wang, A. Sil, and H. Hajishirzi, “Self-RAG: Learning to retrieve, generate, and critique through self-reflection,” in Proc. Int. Conf. Learn. Represent. (ICLR), 2024

    2024

  18. [27]

    Exploring faithful rationale for multi-hop fact verification via salience-aware graph learning,

    J. Si, Y . Zhu, and D. Zhou, “Exploring faithful rationale for multi-hop fact verification via salience-aware graph learning,” in Proc. AAAI Conf. Artif. Intell. (AAAI), 2023, pp. 13573–13581

    2023

  19. [28]

    Answering complex open- domain questions with multi-hop dense retrieval,

    W. Xiong, X. L. Li, S. Iyer, J. Du, P. Lewis, W. Y . Wang, Y . Mehdad, S. Yih, S. Riedel, D. Kiela, and B. Oguz, “Answering complex open- domain questions with multi-hop dense retrieval,” in Proc. Int. Conf. Learn. Represent. (ICLR), 2021

    2021

  20. [29]

    Baleen: Robust multi-hop reasoning at scale via condensed retrieval,

    O. Khattab, C. Potts, and M. A. Zaharia, “Baleen: Robust multi-hop reasoning at scale via condensed retrieval,” in Adv. Neural Inf. Process. Syst. (NeurIPS), 2021, pp. 27670–27682

    2021

  21. [30]

    RECOMP: Improving retrieval-augmented LMs with context compression and selective augmentation,

    F. Xu, W. Shi, and E. Choi, “RECOMP: Improving retrieval-augmented LMs with context compression and selective augmentation,” in Proc. Int. Conf. Learn. Represent. (ICLR), 2024

    2024

  22. [31]

    Causal Walk: Debiasing multi-hop fact verification with front-door adjustment,

    C. Zhang, L. Zhang, and D. Zhou, “Causal Walk: Debiasing multi-hop fact verification with front-door adjustment,” in Proc. AAAI Conf. Artif. Intell. (AAAI), 2024, pp. 19533–19541

    2024

  23. [32]

    Read it twice: Towards faithfully interpretable fact verification by revisiting evidence,

    X. Hu, Z. Hong, Z. Guo, L. Wen, and P. S. Yu, “Read it twice: Towards faithfully interpretable fact verification by revisiting evidence,” in Proc. Int. ACM SIGIR Conf. Res. Develop. Inf. Retrieval (SIGIR), 2023, pp. 2319–2323

    2023

  24. [33]

    LOREN: Logic-regularized reasoning for interpretable fact verification,

    J. Chen, Q. Bao, C. Sun, X. Zhang, J. Chen, H. Zhou, Y . Xiao, and L. Li, “LOREN: Logic-regularized reasoning for interpretable fact verification,” in *Proc. AAAI Conf. Artif. Intell. (AAAI)*, 2022, pp. 10482–10491

    2022

  25. [34]

    GraphFormers: GNN-nested transformers for representation learning on textual graph,

    J. Yang, Z. Liu, S. Xiao, C. Li, D. Lian, S. Agrawal, A. Singh, G. Sun, and X. Xie, “GraphFormers: GNN-nested transformers for representation learning on textual graph,” in *Adv. Neural Inf. Process. Syst. (NeurIPS)*, 2021, pp. 28798–28810

    2021

  26. [36]

    Evidence Retrieval is almost All You Need for Fact Verification,

    L. Zheng, C. Li, X. Zhang, Y . Shang, F. Huang, and H. Jia, “Evidence Retrieval is almost All You Need for Fact Verification,” in *Findings of the Association for Computational Linguistics (ACL)*, 2024, pp. 9274–9281

    2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.