UNI-EXP-002 · Implementation Record
Family Assignment and fixed-seed Sampling Manifests are frozen.
词族分配与固定种子抽样清单已经冻结。
Completed controls
Complete 74,286-word text release downloaded from Ghent University on 2026-08-28.
subtlexus2.zip67e595da…ad83856c5f86f06…29c86e4下载容器与实际读取表分别留存校验和，原始文件不经改名或重存后再计算来源哈希。
Conservative WordNet 3.0 Morphy rules merge only unique supported forms. Ambiguous forms remain isolated from the eligible pool until reviewed.
cbda5ea6…e2646ab59只在 WordNet 支持且结果唯一时合并；多解词形进入独立审查队列，不被强行词元化。
Unconditional binomial enumeration evaluates every possible table with the two-sided Fisher exact conditional p-value.
参考情景的方向一致精确检验功效为 0.858948，达到预定的 0.80 目标；这不保证更小效应或其他基线下的功效。
总体审计与词族审查候选队列完全不使用目标语义标签，不能被解释为最终样本或实验结果。
Frozen manifests
1,080 lemmas assigned to 971 semantic-label-blind structural families.
120 W and 240 Controls; 360 unique Family IDs under seed 20260827002.
Annotator A/B work remains outside this completed implementation phase.
No locked semantic dataset or result exists.
No target semantics or result has been inspected. Implementation status is not evidence for or against the W hypothesis. · 尚未查看目标语义或实验结果；实施状态不构成支持或反对 W 假说的证据。