受治理的演绎：超越相关性的策略基础前提授权

中文摘要 AI 辅助

推理系统通常将前提使用视为相关性问题：如果一个事实可用且有用，它可能被选择用于推理。授权施加了不同的约束：一个前提可能被表示且在逻辑上可用，但未被允许用于特定的局部转换。我们将这一区别形式化为受治理的演绎（GD），并引入转换局部接纳谓词 admit(p, tau, S)。基于独立生成的 RBAC 增强版 Spider 基准，我们构建了 4,461 个匹配的授权对，其中相同的查询前提和策略状态支持允许和禁止的消费转换。初始联合控制器达到 99.19% 的留出准确率，但仅转换控制器达到 100%，暴露了角色名称捷径。在冻结的、标签无关的上下文局部角色置换消除该捷径后，仅前提/状态、仅转换和联合线性控制器在 1,856 条留出边上均恰好得分 50%，而符号策略预言机仍保持 100%。结果是一个受控的负面发现：该基准实例化了超越相关性的策略基础授权，但冻结的线性表示未能恢复该关系。因此，匹配的单侧控制和泄漏审计对于评估学习到的策略敏感推理至关重要。

英文摘要

Reasoning systems usually treat premise use as a question of relevance: if a fact is available and useful, it may be selected for inference. Authorization imposes a different constraint: a premise may be represented and logically usable but not permitted for a particular local transition. We formalize this distinction as Governed Deduction (GD), with a transition-local admission predicate admit(p, tau, S). From an independently produced RBAC-augmented Spider benchmark, we construct 4,461 matched authorization pairs in which the same query premise and policy state support permitted and denied consuming transitions. An initial joint controller reaches 99.19% held-out accuracy, but a transition-only control reaches 100%, exposing a role-name shortcut. After a frozen, label-independent context-local role permutation removes that shortcut, premise/state-only, transition-only, and joint linear controllers all score exactly 50% on 1,856 held-out edges, while a symbolic policy oracle remains at 100%. The result is a controlled negative finding: the benchmark instantiates policy-grounded authorization beyond relevance, but the frozen linear representation does not recover the relation. Matched one-sided controls and leakage audits are therefore essential for evaluating learned policy-sensitive reasoning.

发表机构

• The Institute of Energetic Paradigm（能量范式研究所）
• National Center for High-Performance Computing, Taiwan（台湾国家高速网络与计算中心）

机构由 AI 辅助整理，请以论文原文为准。