← 研踪 PaperTrailarXiv:2611.02775 · v2
Human-Centred, Risk-Constrained Evaluation of Financial Document Recognition: Evidence, Critical Errors, and Selective Automation
首次提交 2026-11-02 16:51
版本更新 2026-11-05 18:23
雷达观测 2026-11-06 12:09
摘要
Financial document recognition is often assessed using character and word error rates, although these metrics do not express whether a system preserves the financial facts on which people and organisations act. This paper develops a human-centred, risk-constrained framework for evaluating financial document recognition as evidence recovery and decision support. It defines typed financial facts linked to visible source regions and separates verbatim transcription, deterministic normalization, and business interpretation. A layered quality model covers text, localization, layout, tables, fields, fact correctness, evidence grounding, uncertainty, robustness, and operational performance. Non-compensatory acceptance gates use one-sided uncertainty bounds for critical-field errors, preventing strong aggregate scores from masking failures in amounts, identifiers, signs, periods, or entity associations. Selective automation is formulated as a constrained routing problem that accounts for residual risk, human-review capacity, and review cost. The framework further specifies annotation and adjudication procedures, leakage-resistant benchmark partitions, statistical estimands, subgroup analyses, and a reproducibility card. Formal results establish the limitations of compensatory scoring, the non-zero upper risk bound after zero observed errors, and a loss sensitive decision rule for routing uncertain cases to human review. The contribution is theoretical and methodological; no new model, dataset, or empirical performance result is claimed.
观测到的版本
相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。
v2 · 2026-11-05 18:23 · 最新观测版本
Human-Centred, Risk-Constrained Evaluation of Financial Document Recognition: Evidence, Critical Errors, and Selective Automation
Financial document recognition is often assessed using character and word error rates, although these metrics do not express whether a system preserves the financial facts on which people and organisations act. This paper develops a human-centred, risk-constrained framework for evaluating financial document recognition as evidence recovery and decision support. It defines typed financial facts linked to visible source regions and separates verbatim transcription, deterministic normalization, and business interpretation. A layered quality model covers text, localization, layout, tables, fields, fact correctness, evidence grounding, uncertainty, robustness, and operational performance. Non-compensatory acceptance gates use one-sided uncertainty bounds for critical-field errors, preventing strong aggregate scores from masking failures in amounts, identifiers, signs, periods, or entity associations. Selective automation is formulated as a constrained routing problem that accounts for residual risk, human-review capacity, and review cost. The framework further specifies annotation and adjudication procedures, leakage-resistant benchmark partitions, statistical estimands, subgroup analyses, and a reproducibility card. Formal results establish the limitations of compensatory scoring, the non-zero upper risk bound after zero observed errors, and a loss sensitive decision rule for routing uncertain cases to human review. The contribution is theoretical and methodological; no new model, dataset, or empirical performance result is claimed.
查看此版本解读 →查看此版本原文 ↗v1 · 2026-11-02 16:51
Risk-Constrained Evaluation of Financial Document Recognition: A Formal Framework for Evidence, Critical Errors, and Selective Automation
Financial document recognition is commonly evaluated using character and word error rates, which do not adequately represent the consequences of errors in financial workflows. This paper formalizes financial document recognition as a risk-constrained evidence-recovery problem. It defines typed, evidence-linked financial facts and a multi-layer quality model spanning text, layout, tables, fields, semantic relations, calibration, robustness, and operational performance. It introduces non-compensatory acceptance gates based on upper confidence bounds for critical-field errors, and formulates selective automation as a constrained risk-coverage problem with review-capacity limits. The paper also specifies annotation requirements, leakage-resistant benchmark splits, statistical estimands, and a reproducibility card. Formal propositions establish why aggregate scores cannot certify critical-field reliability and why zero observed errors do not imply zero risk. A synthetic example illustrates the decision rule. The work is theoretical; no empirical benchmark or model performance claim is presented.
查看此版本解读 →查看此版本原文 ↗