Skip to main content
Aggregate arXiv cs.AI 人工智能 3 Sep 2026 - 12:00

When Can a Machine Trust a Statute? A Survival Certificate for Machine-Extracted Legal Logic

RSS 官方收录 · 可信分层展示

关键摘要

arXiv:2609.…

  • 01741v1 Announce Type: new Abstract: Statutes are increasingly parsed …
  • We ask what formal logic survives such noise.
  • We build a passive survival certificate for the Duquenne-Guigues impli…

摘要引擎:抽取

正文提要

arXiv:2609.01741v1 Announce Type: new Abstract: Statutes are increasingly parsed by machines before people read them, and the parsers disagree: on Missouri's statutes, two independently written extractors diverge on numeric-threshold presence at a false-negative rate of 0.43. We ask what formal logic survives such noise. We build a passive survival certificate for the Duquenne-Guigues implication basis of machine-extracted statutory contexts: per-attribute inter-extractor disagreement is measured, replayed against the basis in 1,000 Monte Carlo trials, and an implication is certified only when a one-sided Wilson 95% lower bound on survival reaches 0.95; every certified implication carries premise spans and a minimal counterexample. On 29,365 Missouri sections and 502 Indian central-Act sections, the preregistered held-out gate passes (10 statute families across 7 Titles exact; 16 across 11 with 5% tolerance), yet under one globally deployed error model 93.2% of held-out chapters fall below the informativeness floor, and a 2x2 factorial assigns that to calibration-rate transfer, not selection. The certificate is usable but fragile: deploy it per-chapter-calibrated or error-tolerant. Code, data products, and the audit trail, including one retracted claim, are released.

来源:https://arxiv.org/abs/2609.01741

打开官方原文 站点原文页 可信分区 本信源更多 今日简报 分享图 RSS 稍后再看列表