Metraipolis › archive › dc_fd3c07465ede

ScienceJudge: Pattern Perturbation Augmentation for Sentence-Level Scientific Claim Verification — arXiv:2610.40238

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.40238
License (as archived)
CC0-1.0
Archived
2026-10-31T06:17:06.244Z
SHA-256 of the archived text
c289a9e7f319fe27daf851233f80c411d8a94ffe7934ee40a072f68ed48a02fc
Stable URL of this page
https://metraipolis.com/documento/dc_fd3c07465ede

How to cite

ScienceJudge: Pattern Perturbation Augmentation for Sentence-Level Scientific Claim Verification — arXiv:2610.40238 — arXiv (en), https://arxiv.org/abs/2610.40238 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fd3c07465ede (Metraipolis archive dc_fd3c07465ede, sha256 c289a9e7f319fe27daf851233f80c411d8a94ffe7934ee40a072f68ed48a02fc)
BibTeX
@misc{dcfd3c07465ede,
  title = {ScienceJudge: Pattern Perturbation Augmentation for Sentence-Level Scientific Claim Verification — arXiv:2610.40238},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.40238},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fd3c07465ede; sha256 c289a9e7f319fe27daf851233f80c411d8a94ffe7934ee40a072f68ed48a02fc},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.40238. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Sentence-level scientific claim verification determines whether each sentence in an AI-generated scientific paragraph is supported by a given multimodal evidence bundle. Fine-grained reporting errors often preserve fluent and plausible statements while introducing subtle evidence-claim deviations, making them difficult to detect under sparse and imbalanced supervision. We propose Pattern Perturbation Augmentation (PPA), which models reporting errors as label-specific faithfulness breaking transformations and induces reusable error patterns from labeled examples. PPA applies these patterns to supported claims to construct controlled hard negatives and expands their coverage through pattern-level, operation-level, and cross-domain augmentation. Based on these augmented views, we develop ScienceJudge, which integrates PPA-trained vision-language predictors and instruction-guided LLM predictors through class-weighted ensemble inference. On the local development set, ScienceJudge improves the average score from 0.8124 to 0.8692, achieving a Macro-F1 of 0.8466 and a Paragraph Exact Match score of 0.8919. Our submitted system ranks first on the official NLPCC 2026 Task 10 Track 1 leaderboard with a final score of 0.6890.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2610.40238 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-31T06:07:23.370Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).