Pith. sign in

REVIEW 3 major objections 5 minor 29 references

ProvenanceGuard: Source-Aware Factuality Verification for MCP-Based LLM Agents

T0 review · 3 major / 5 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read Source attribution is an independent axis of factuality for MCP-grounded agents: a supported claim can still be wrong if credited to the wrong evidence source.

desk verdict A well-scoped, honestly limited systems paper; the task formulation is the contribution, and the headline numbers need independent confirmation before being treated as established. read the letter →

arxiv 2606.18037 v2 pith:ZESHNARJ submitted 2026-06-16 cs.AI cs.CLcs.MA

Ander Alvarez , Santhiya Rajan , Samuel Mugel , Román Orús This is my paper · ORCID
classification cs.AIcs.CLcs.MA
keywords sourceattributionfactualityverificationcross-sourceconflationMCPagentsprovenanceclaimdecompositionNLIsupportchecking
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

Tool-using AI agents that gather evidence through MCP servers (a standard way of connecting agents to external tools) often mix several sources into one answer. This paper claims that current factuality checks miss a specific failure: a claim can be supported by some piece of evidence while being attributed to the wrong source. It introduces ProvenanceGuard, which routes each atomic claim to the source that actually supports it, then checks whether the answer named that source; answers that conflate sources are blocked. On a held-out set of medical agent traces the verifier matches source-blind support checkers while also reporting claim-to-source attribution, and it catches all 50 deliberately injected attribution swaps. If this holds, factuality evaluation for tool-using agents needs a source-ownership axis, not just a pooled-support score.

What carries the argument

The load-bearing mechanism is the separation of support from source ownership. The verifier preserves stable MCP tool IDs and source IDs, routes each claim to the highest-scoring evidence source via centroid cosine similarity, scores support with NLI plus a token-alignment proxy and protected-value checks, and then — only after the routed source is fixed — compares the claim's stated or implied attribution with that routed source. This two-step split turns cross-source conflation into a detectable verdict instead of an invisible failure. A random-forest calibration layer sets the conservative support threshold, and a fail-closed aggregation rule blocks an answer if any claim fails either sup

What would settle it

Have clinicians annotate the training and validation claims from the same traces (or re-annotate the held-out 361) without seeing the system's verdicts; if source accuracy on this human-only gold set falls well below 0.858, or if the 50/50 conflation results do not survive paraphrased or multi-swap probes, the central claim would need qualification.

Watch

Extended reading notes

Core claim

The core discovery is that source attribution is a separate and measurable axis of factuality in MCP-grounded answers. A claim may be entailed by pooled evidence and still be wrong because the answer assigns it to the wrong tool output, patient record, literature source, or metadata source — the pattern the paper calls cross-source conflation. ProvenanceGuard makes this distinction operational by carrying stable source IDs through claim decomposition, source routing, NLI-based support checking with token-alignment and protected-value checks, and a calibrated support decision; only then does it compare stated attribution with the routed supporting source. The paper reports held-out block F1 0

Load-bearing premise

The headline numbers rest on labels that are mostly LLM-assisted; only the held-out 361 labels got human review, so if the judge model and the verifier share the same blind spots about what counts as source support, the reported source accuracy could be self-confirming.

Editorial extensions

If this is right

  • Factuality evaluation for tool-using agents should report source ownership separately from pooled support; source-blind scores cannot detect cross-source conflation.
  • A verifier that preserves stable source IDs can match source-blind support performance (held-out block F1 0.802) while also emitting claim-to-source attribution.
  • Exact source ownership is substantially harder than support detection when candidate sources are semantically close: source-plus-relation accuracy drops to 0.229 on the harder multi-source benchmark.
  • Repair-and-reverify can turn every blocked answer into a verifier-passing answer, but in most full-trace cases the repair is a conservative fallback rather than a substantive rewrite.

Reading between the lines

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

  • Beyond the paper: if source attribution is a separate axis, evaluation suites for tool-using agents should include deliberately confused-source cases (e.g., a fact supported by the chart but attributed to literature) as a standard stress slice, not just incidental trace data.
  • Beyond the paper: the reported 50/50 conflation detections are on clean single-swap probes; a natural next test is multi-error, paraphrased, or adversarially hidden swaps, where precision is unknown.
  • Beyond the paper: because only held-out labels were human-reviewed, an independent clinician gold set over the training and validation claims would tell whether the calibration and source-accuracy numbers are stable or partly self-confirming.
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

3 major / 5 minor

Summary. The paper introduces ProvenanceGuard, a source-aware factuality verifier for MCP-based LLM agents. It decomposes an agent answer into atomic claims, routes each claim to a specific MCP evidence source using embeddings, checks support with NLI plus a token-alignment and protected-value heuristic, calibrates the support decision with a random-forest model, and then compares the routed source against the source stated or implied in the answer. The system returns per-claim verdicts and an allow/block answer decision, with a RARR-style repair loop for blocked answers. Evaluation uses a frozen corpus of 281 captured medical-domain MCP traces; the primary held-out split is 40 traces / 361 human-reviewed claims, with reported block F1 0.802 and source accuracy 0.858 over 260 source-eligible claims. A harder multi-source benchmark yields block F1 0.846 but source-plus-relation accuracy 0.229, and 50 controlled single-swap conflation probes are all blocked and repaired. The paper argues that source attribution is an independent axis of factuality verification beyond pooled evidence support.

Significance. If the central claim holds, the paper addresses a real and underexplored failure mode: a claim can be supported by pooled evidence while being attributed to the wrong MCP tool output. The system's preservation of stable source IDs through routing and support checking is a sensible design, and the comparison with source-blind baselines (MiniCheck, RAGAS, AlignScore, SummaC-ZS) is appropriate. The paper is also commendably honest about limitations: it reports wide confidence intervals, acknowledges that human review covers only the held-out labels, separates the multi-source benchmark units, and does not overclaim the repair loop as an independent oracle. However, the load-bearing evidence for the source-attribution claim rests on (i) a gold standard whose human-review protocol is not described and whose training/validation labels are LLM-assisted only, and (ii) 50 conflation probes that are deliberately simple single-swap family-level cases. The multi-source benchmark, with source-plus-relation accuracy of only 0.229, further tempers the strength of the empirical support for exact source ownership. The contribution is therefore better characterized as a well-engineered demonstration

major comments (3)
  1. [Sections IV-H, V-G, VIII; Table XXI] The gold standard for the headline source metrics is not shown to be independent of the verifier. Human review covers only the 361 held-out labels; training and validation labels are LLM-assisted, and the human protocol is unspecified: no inter-annotator agreement, no statement about blinding to model labels, no report of how often reviewers changed labels, and no evidence that reviewers saw complete raw tool outputs. If the two Gemma judge prompts share the same semantic-similarity notion as the MiniLM router and DeBERTa NLI, source accuracy 0.858 and block F1 0.802 could partly reflect agreement with the judge's relevance heuristic rather than true provenance. This is the primary evidence for the central claim. Please provide a detailed re-annotation protocol and/or an independent human gold set for at least a random sample, with agreement statistics and a list of label changes.
  2. [Section V-F and Section VIII] The 50/50 conflation result is the only direct evidence for cross-source conflation detection, but each probe is a single family-level swap (chart vs. literature) with no paraphrase, no adversarial hiding, and no near-source confound. The lexical alias matcher and routing are designed exactly for this contrast, so the exact binomial interval [0.93, 1.00] supports only the narrow claim that simple attribution swaps are blocked. The central novelty claim — that cross-source conflation is an independent failure mode a verifier can catch — would be much stronger with probes involving semantically close sources, paraphrased attribution spans, or multi-error traces. The conclusion should not rely on the 50/50 number as evidence beyond the simple-swap setting.
  3. [Sections III-H and V-E; Table XIX] Repair-and-reverification success is scored by the same verifier that issued the block, so the abstract's statement that the loop 'resolves all blocked answers' demonstrates only pipeline consistency, not independent correctness. Table XIX shows 144 of 173 resolutions are terminal conservative fallbacks rather than substantive answer recoveries. The paper acknowledges this in the discussion, but the abstract and contribution list overstate the repair result. Please either measure repaired answers under independent human or LLM adjudication, or rephrase the claim as 'all blocked answers are converted to verifier-passing outputs, mostly by conservative fallback.'
minor comments (5)
  1. [Section V-A; Table III] The held-out block-F1 confidence interval [0.664, 0.900] is very wide, and the MiniCheck comparison in Table XVII is not statistically significant (one-sided p≈0.13). The text handles this adequately, but the abstract could state more explicitly that the support-F1 advantage over source-blind baselines is not the main evidence for the paper's contribution.
  2. [Section V-B; Table VIII] Decomposition is measured against a frozen reference extraction, not human gold, and the protected-value exact rate of 0.563 is low. This is acknowledged in the limitations, but it deserves a more prominent place in the answer-level repair discussion since protected-value errors can affect downstream attribution decisions.
  3. [Section V-B; Table XIV] The 2048-token ModernBERT diagnostic is a one-epoch, 100-group replacement checkpoint with raw source-plus-relation accuracy 0.090. Reporting it as a 'valid raw test point' is fine, but the row reads as if it were a substantive long-context result; consider moving this diagnostic clearly to an appendix or supplement.
  4. [Section IV-C] The unit mismatch between 254 pairwise claim cases and 263 frozen extracted claims for the same 59 test questions is clearly stated, but the narrative occasionally refers to both as 'benchmark results.' Please ensure each table and sentence explicitly identifies which unit is being used.
  5. [Section II] Minor typo: 'F ActScore' should likely be 'FactScore'. Also, the related-work discussion of ALCE is helpful and should be kept, but it could note more explicitly that ALCE's citation-level attribution is a passage-level task, not a tool-provenance task.

Circularity Check

1 steps flagged · score 4.0 of 10

Repair-and-reverify success reduces to the same verifier's own verdict; terminal fallback makes 'all resolved' near-tautological. The central held-out support/source metrics remain independent.

  1. self definitional [Section III-H (Repair and Reverification); Section V-E (RQ4); Table XIX]
    "The RARR-style repair loop is therefore not treated as a separate oracle; it is a repair mechanism whose output must satisfy the same attribution-sensitive checks that blocked the original answer. ... The limitation is that repair success is measured against the same verifier that triggered the block. ... The repair loop resolves all 173 blocked answers, and all 173 revised outputs pass the same Router+NLI verifier."

    A blocked answer counts as 'resolved' exactly when the same ProvenanceGuard verifier passes the revised output, and the loop is allowed to terminate by replacing remaining content with a conservative non-claim response. Because the verifier treats generic safety boilerplate as non-evidence-bearing, the terminal fallback path (144 of 173 resolutions in Table XIX) passes by construction. Thus 'repair resolves all blocked answers' is not a prediction about external correctness; it is an artifact of defining success as the verifier's own verdict. The paper concedes this is 'not independent proof,' but the headline repair/resolution numbers still reduce to the verifier's self-judgment rather than to an external gold standard.

full rationale

The core derivation is not circular. The claim-level support and source-attribution numbers are measured on a held-out packet with human-reviewed labels (Section IV-H, Table XXI), the router's 0.858 source accuracy is a Top-1 cosine ranking compared against those labels, and the source-blind baselines are external systems never used to train or calibrate ProvenanceGuard. I found no self-citation chain, no imported uniqueness theorem, and no fitted parameter renamed as a prediction. The main circular element is the repair-and-reverify evaluation: 'resolved' is defined as passing the same verifier that produced the block, and the terminal fallback rule makes passing near-guaranteed. The paper explicitly acknowledges this limitation ('repair success is measured against the same verifier that triggered the block... not independent proof'), so it is a transparently disclosed partial circularity rather than a hidden one. The 50 conflation probes are controlled diagnostics whose ground truth matches the system's own attribution rule; I did not count this as a separate circular step because the probes still require routing and NLI to succeed, but their evidentiary weight is limited. Overall: the central claim that source attribution is an independent axis has independent support from the held-out source-accuracy and multi-source benchmark; the repair claims are partly self-confirming, warranting a score of 4 rather than 0.

Assumptions & free parameters 9 free parameters · 6 assumptions · 0 invented entities

The central claim rests on the MCP trace interface preserving source identity, on the routing/NLI stack being competent, and on the adjudicated labels being trustworthy. No new physical or conceptual entities are introduced. More than a dozen hand-set thresholds and model choices are fitted or tuned on development/validation data; the support threshold alone moves the operating point substantially.

free parameters (9)
  • support threshold = 0.65
    Selected on validation by block F1 (Table I, §III-E); directly sets allow/block operating point.
  • alignment threshold tau = 0.35 of maximum claim-token attention score
    Hand-set threshold marking weakly grounded tokens (§III-D, Table II).
  • minimum supported-token ratio = 0.70
    Hand-set ratio for entailment claims with token alignment (§III-D, Table II).
  • lexical rescue thresholds = 0.55 (with protected values); 0.85 (otherwise)
    Hand-set thresholds controlling when neutral NLI is lexically rescued (§III-D, Table II).
  • NLI pair budget = 512 tokens (previously 256)
    Truncation budget for premise-claim pair; changed during development (§IV-D, Table II).
  • random forest hyperparameters = 400 trees, depth 5, min leaf 8, balanced class weights, seed 20260607
    Chosen configuration for calibrated support classifier (§III-E, Table I).
  • max extracted claims M = 20
    Cap on atomic claims per answer (§III-B, Table II).
  • source embedding model = all-MiniLM-L6-v2
    Choice of embedding model for source routing; affects routing scores and margins (Table II).
  • NLI checkpoint = MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli
    Pretrained NLI model used as support scorer; not fine-tuned here (Table II).
assumptions (6)
  • domain assumption MCP trace objects carry stable source IDs and complete raw tool outputs; if no source ID, tool name fallback suffices.
    Interface definition in §III-A; central to preserving provenance.
  • domain assumption The cosine-similarity centroid router ranks the true supporting source first often enough; routing errors cascade to all downstream verdicts.
    §III-C; router-only top-1 accuracy is 0.858 on the easy held-out split but lower on the multi-source benchmark.
  • domain assumption The NLI model and attention-token alignment proxy measure support for the routed source.
    §III-D; no independent validation of NLI reliability on MCP evidence.
  • domain assumption The LLM-assisted adjudication labels (Gemma 4 E4B judge prompts) and 361 human-reviewed labels are valid gold.
    §IV-H and §V-G; training/validation labels are not human-reviewed.
  • ad hoc to paper Passing the same verifier after repair is evidence of resolution.
    §V-E; paper explicitly says this is pipeline consistency, not independent clinical validation.
  • domain assumption Fail-closed blocking (block if any claim fails) is the right evaluation policy.
    §III-G; policy choice affects precision/recall tradeoff.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ProvenanceGuard: Source-Aware Factuality Verification for MCP-Based LLM Agents." pith.science (2026). https://pith.science/paper/ZESHNARJ

@misc{pith2026260618037,
  author       = {Pith},
  title        = {Pith review of: ProvenanceGuard: Source-Aware Factuality Verification for MCP-Based LLM Agents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZESHNARJ}},
  note         = {Machine review of arXiv:2606.18037}
}
read the original abstract

Tool-using LLM agents increasingly use the Model Context Protocol (MCP) to answer from heterogeneous evidence sources, including search, APIs, databases, clinical records, and formulary tools. Standard factuality metrics usually test whether an answer is supported by pooled evidence, missing a provenance-sensitive failure mode: a claim may be supported somewhere while being attributed to the wrong source. We call this cross-source conflation. We introduce ProvenanceGuard, a source-aware verifier for MCP-grounded answers. It consumes captured MCP traces with stable tool IDs, source IDs, and raw outputs; decomposes answers into atomic claims; routes claims to source-specific evidence; checks support with NLI and a token-alignment proxy; compares stated attribution with the routed source; and returns per-claim verdicts plus an answer-level allow/block decision. Blocked answers can be repaired with retrieval-augmented answer revision and re-verified. We evaluate on 281 medical-domain MCP-agent traces. A 266-trace adjudicated subset yields 2,325 LLM-assisted claim labels split by trace; 361 held-out labels are human-verified. On the 40-trace held-out split, ProvenanceGuard achieves block F1 0.802 and source accuracy 0.858 over 260 source-eligible claims, outperforming source-blind baselines that do not emit claim-to-source IDs. On a harder multi-source benchmark it reaches block F1 0.846, while source-plus-relation accuracy drops to 0.229, showing that exact source ownership remains difficult with semantically close sources. Repair-and-reverify resolves all blocked answers in the full trace set, often via conservative fallback. In 50 controlled clinical conflation probes, ProvenanceGuard detects all injected attribution swaps with no retained wrong attribution. These results show that source attribution is an independent axis for factuality verification in MCP-based agents.

Figures

Figures reproduced from arXiv: 2606.18037 by the authors.

Figure 1
Figure 1. FIG. 1. Why source-aware factuality is stricter than source-blind support. A claim can be supported by one MCP source while [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Sequential source-aware verification pipeline. The agent core calls MCP tools and produces a draft answer; [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Calibration layer. The calibrator receives only verifier-internal routing, NLI, lexical, token-alignment, and protected [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: FIG. 4. Evaluation datasets and units used in the paper. The primary captured-trace corpus supports claim-level scoring, [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 3 linked inside Pith

  1. [1]

    S. Min, K. Krishna, X. Lyu, M. Lewis, W.-t. Yih, P. W. Koh, M. Iyyer, L. Zettlemoyer, and H. Hajishirzi, inPro- ceedings of the 2023 Conference on Empirical Methods in 17 System Claims Block P Block R Block F1 Verdict acc. Verdict macro F1 ProvenanceGuard (ours) 361 0.673 0.993 0.802 0.812 0.406 MiniCheck 361 0.655 0.971 0.783 0.792 0.396 RAGAS Faithfulne...

    2023

  2. [2]

    We introduceProvenanceGuard, a calibrated source-aware Router+NLI verifier that decomposes answers into claims, preserves stable MCP tool IDs and source IDs from raw tool outputs, routes claims to source-specific evidence, and detects cross-source conflation

  3. [3]

    We report a frozen captured medical-domain agent benchmark built from 281 captured MCP-agent traces, a 266-trace claim subset with 2,325 LLM- assisted labels, and a 40-trace held-out packet with complete MCP tool outputs and human-expert- reviewed labels

  4. [4]

    We compareProvenanceGuardagainst MiniCheck, RAGAS Faithfulness, AlignScore, and SummaC-ZS on the same held-out claim packet, while separately evaluating RARR-style repair on full captured traces and targeted source-conflation probes. The central claim is limited to source-attribution factu- ality in MCP-grounded answers; we do not claim to solve open-doma...

  5. [5]

    Y. Song, Y. Kim, and M. Iyyer, inFindings of the Asso- ciation for Computational Linguistics: EMNLP(2024)

    2024

  6. [6]

    Goyal and G

    T. Goyal and G. Durrett, inFindings of the Association for Computational Linguistics: EMNLP 2020(2020)

    2020

  7. [7]

    L. Tang, P. Laban, and G. Durrett, inProceedings of the 2024 Conference on Empirical Methods in Natural Lan- guage Processing(Association for Computational Lin- guistics, Miami, Florida, USA, 2024) pp. 8818–8847

    2024

  8. [8]

    Laban, T

    P. Laban, T. Schnabel, P. N. Bennett, and M. A. Hearst, SummaC: Re-visiting nli-based models for inconsistency detection in summarization (2021), arXiv:2111.09525

    arXiv 2021

Show all 29 references
  1. [9]

    Y. Zha, Y. Yang, R. Li, and Z. Hu, inProceedings of the 61st Annual Meeting of the Association for Compu- tational Linguistics(Association for Computational Lin- guistics, Toronto, Canada, 2023) pp. 11328–11348

    2023

  2. [10]

    evaluates whether LLM-generated citations point to the correct supporting passage, which is the closest existing task to source attribution. However, ALCE operates at the passage or chunk level within a single retrieved set, whereas MCP traces expose stable tool- level source ...

  3. [11]

    Bohnet, V

    B. Bohnet, V. Q. Tran, P. Verga, R. Aharoni, D. Andor, L. B. Soares, M. Ciaramita, J. Eisenstein, K. Ganchev, J. Herzig, K. Hui, T. Kwiatkowski, J. Ma, J. Ni, L. Sestorain Saralegui, T. Schuster, W. W. Cohen, M. Collins, D. Das, D. Metzler, S. Petrov, and K. Web- ster, Attribu...

    2022 arXiv

  4. [12]

    Cattan, P

    A. Cattan, P. Roit, S. Zhang, D. Wan, R. Aharoni, I. Szpektor, M. Bansal, and I. Dagan, Transactions of the Association for Computational Linguistics (2026)

    2026

  5. [13]

    Harary, E

    S. Harary, E. Hirsch, A. Slobodkin, D. Wan, M. Bansal, and I. Dagan, PrefixNLI: Detecting factual inconsisten- cies as soon as they arise (2025), arXiv:2511.01359

    2025

  6. [14]

    S. Es, J. James, L. Espinosa-Anke, and S. Schockaert, inProceedings of the 18th Conference of the European Chapter of the Association for Computational Linguis- tics: System Demonstrations(Association for Computa- tional Linguistics, St. Julian’s, Malta, 2024) pp. 150–158

    2024

  7. [15]

    T. Gao, H. Yen, J. Yu, and D. Chen, inProceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP)(Association for Compu- tational Linguistics, Singapore, 2023) pp. 10355–10377

    2023

  8. [16]

    D. Wan, E. Hirsch, E. Stengel-Eskin, I. Dagan, and M. Bansal, GenerationPrograms: Fine-grained attribu- tion with executable programs (2025), arXiv:2506.14580

    2025 arXiv

  9. [17]

    X. Yue, B. Wang, Z. Chen, K. Zhang, Y. Su, and H. Sun, inFindings of the Association for Computational Lin- guistics: EMNLP 2023(Association for Computational Linguistics, Singapore, 2023) pp. 4615–4635

    2023

  10. [18]

    Rashkin, V

    H. Rashkin, V. Nikolaev, M. Lamm, L. Aroyo, M. Collins, D. Das, S. Petrov, G. S. Tomar, I. Turc, and D. Reitter, Computational Linguistics49, 777 (2023)

    2023

  11. [19]

    Honovich, R

    O. Honovich, R. Aharoni, J. Herzig, H. Taitelbaum, D. Kukliansy, V. Cohen, T. Scialom, I. Szpektor, A. Has- sidim, and Y. Matias, inProceedings of the 2022 Confer- ence of the North American Chapter of the Association 18 for Computational Linguistics: Human Language Tech- nolo...

    2022

  12. [20]

    takes a generated passage, researches evidence, and revises unsupported claims while preserving the original style and structure. In our setting, RARR-style repair is evaluated after source-aware blocking: the verifier rejects an answer, repair attempts to produce a source-gro...

  13. [21]

    Hirsch, A

    E. Hirsch, A. Slobodkin, D. Wan, E. Stengel-Eskin, M. Bansal, and I. Dagan, inProceedings of the 63rd An- nual Meeting of the Association for Computational Lin- guistics(2025) pp. 15355–15370

    2025

  14. [22]

    J. Gao, J. Zhou, Q. Sun, R. Huang, and S. Yoo, Atomic information flow: A network flow model for tool attribu- tions in RAG systems (2026), arXiv:2602.04912

    2026

  15. [23]

    Zhang, Z

    Q. Zhang, Z. Xiang, Y. Xiao, L. Wang, J. Li, X. Wang, and J. Su, inProceedings of the 63rd Annual Meeting of the Association for Computational Linguistics(2025)

    2025

  16. [24]

    Filice, E

    S. Filice, E. Haramaty, G. Horowitz, Z. Karnin, L. Lewin- Eytan, and A. Shtoff, inProceedings of the 14th Interna- tional Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics(2025) p...

    2025

  17. [25]

    L. Gao, Z. Dai, P. Pasupat, A. Chen, A. T. Chaganty, Y. Fan, V. Y. Zhao, N. Lao, H. Lee, D.-C. Juan, and K. Guu, inProceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)(Association for Computational Linguis- tics, Toron...

    2023

  18. [26]

    P. Roit, J. Ferret, L. Shani, R. Aharoni, G. Cideron, R. Dadashi, M. Geist, S. Girgin, L. Hussenot, O. Keller, N. Momchev, S. Ramos Garea, P. Stanczyk, N. Vieillard, O. Bachem, G. Elidan, A. Hassidim, O. Pietquin, and I. Szpektor, inProceedings of the 61st Annual Meeting of th...

    2023

  19. [27]

    Kry´ sci´ nski, B

    W. Kry´ sci´ nski, B. McCann, C. Xiong, and R. Socher, in Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)(Association for Computational Linguistics, 2020) pp. 9332–9346

    2020

  20. [28]

    I. Poey, J. Liu, and Q. Zhong, inProceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: Industry Track(2025) pp. 1057–1071

    2025

  21. [29]

    question_id

    ThoughtProof, Reasoning verification capability — veri- fying ai output correctness through MCP, MCP GitHub Discussion #2574 (2026). 19 Appendix A: Illustrative Escaped JSON Schema F ragments { "question_id": "clin-real-v2-shuf-003", "user_question": "Summarize beta blockers i...

    2026

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.