Metraipolis › archive › dc_5c69b402e49d
When Self-Verification Hurts: Calibrated Answer Repair for Financial Question Answering — arXiv:2611.05166
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.05166
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-04T12:33:35.796Z
- SHA-256 of the archived text
daf23b3e288f95b05bf53e75017292d029b15966b0d1d9dda9fcbd62e24f9582- Stable URL of this page
https://metraipolis.com/documento/dc_5c69b402e49d
How to cite
When Self-Verification Hurts: Calibrated Answer Repair for Financial Question Answering — arXiv:2611.05166 — arXiv (en), https://arxiv.org/abs/2611.05166 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5c69b402e49d (Metraipolis archive dc_5c69b402e49d, sha256 daf23b3e288f95b05bf53e75017292d029b15966b0d1d9dda9fcbd62e24f9582)
BibTeX
@misc{dc5c69b402e49d,
title = {When Self-Verification Hurts: Calibrated Answer Repair for Financial Question Answering — arXiv:2611.05166},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.05166},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5c69b402e49d; sha256 daf23b3e288f95b05bf53e75017292d029b15966b0d1d9dda9fcbd62e24f9582},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.05166. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Financial question answering can fail even when a model extracts a plausible number from the supplied evidence: the answer may attach it to the wrong entity, period, unit, or accounting concept. Asking the model to verify and revise its draft can help, but open-ended revision may alter content that was already correct or inspect the wrong relation and reinforce the original error. We formulate repair as a decision that balances the information it may recover against the correct content it may damage. Based on this trade-off, we introduce Calibrated Answer Repair (CAR), which first estimates whether a draft is likely to benefit from repair, then selects a targeted check, localizes the relevant evidence or draft span, and limits the permitted edit. Across FinRank, Fin-RATE, and FinQA with three open-weight models, CAR outperforms generic verification in all dataset-model settings. At the system level, it recovers 14.6% of initially missed answer requirements while damaging 2.1% of initially correct ones. Its post-draft token cost is over 30 percent lower than the iterative revision alternatives. Ablations on two dataset-model settings show that evidence localization and bounded editing are the main contributors to CAR's gains. 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/2611.05166 — 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 2026-11-04T12:27:02.054Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).