메뉴
当你在为《自动化学报》(Acta Automatica Sinica, AAS) 提升实验可复现性、组织复现包、防范数据污染时调用。覆盖环境与依赖固定、随机种子与多次运行、数据划分与防泄露、控制仿真的求解器/步长/系统参数记录、超参数与调参协议披露、以及配合双盲的匿名化复现材料，帮助控制/自动化/模式识别方向的中文稿在 Acta Automatica Sinica 外审中经得起"结果能否复现"的追问。
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when preparing the accountability artifacts that accompany an ACM FAccT paper — datasheets for datasets, model cards, data statements, audit and impact-assessment documentation, and released code/data — since FAccT's culture centers documentation and accountability infrastructure rather than a formal ACM artifact-badging track; covers what makes each genre credible, anonymized-review versus public-release versions, and consistency with the paper's harm claims.
Use when drafting ACM FAccT author responses — the short factual-correction rebuttal after preliminary reviews, and, distinctively for the 2026-new process, the Revise-and-resubmit round where you address Area-Chair-prioritized concerns from a mixed CS+law+social-science panel, map each to a concrete change, and survive a re-review while keeping mutual anonymity.
Use when preparing an accepted ACM FAccT paper for camera-ready — de-anonymizing and adding the withheld Positionality/Acknowledgements statements, switching to the acmart sigconf proceedings format, ACM metadata (DOI, ORCID, CCS concepts, rights), the two-round camera-ready calendar for Accept vs Revise papers, ACM Open Access, finalizing datasheets/model cards, and integrating required changes without scope creep.
Use when designing or auditing ACM FAccT empirical work — quantitative fairness audits with disaggregated metrics and fair baselines, qualitative and participatory studies with coding and reflexivity, mixed-methods designs, sound handling of protected attributes and proxies, consent and IRB for human-subjects and community-facing work, and matching evidence to the shape of a fairness/accountability/transparency claim.
Use when positioning an ACM FAccT submission across its many disciplinary lanes — algorithmic fairness/ML, HCI, law and policy, STS and critical theory, and prior FAccT/FAT* proceedings — writing delta-first contrast that a mixed reviewer pool will accept, citing borrowed constructs to their real origin, keeping self-citations mutually anonymous, and declaring overlap with workshops, preprints, and prior versions.
Use when strengthening ACM FAccT transparency and reproducibility — releasing code, data, and analysis for quantitative audits; documenting datasets and models with datasheets, model cards, and data statements; making qualitative and participatory work auditable without breaking confidentiality; pinning provenance for scraped and model-generated data; and keeping the paper, the supplementary material, and any released artifact consistent.
| name | aas-reproducibility |
| description | 当你在为《自动化学报》(Acta Automatica Sinica, AAS) 提升实验可复现性、组织复现包、防范数据污染时调用。覆盖环境与依赖固定、随机种子与多次运行、数据划分与防泄露、控制仿真的求解器/步长/系统参数记录、超参数与调参协议披露、以及配合双盲的匿名化复现材料，帮助控制/自动化/模式识别方向的中文稿在 Acta Automatica Sinica 外审中经得起"结果能否复现"的追问。 |
本技能帮你把投向《自动化学报》(Acta Automatica Sinica, AAS) 的实验做到可复现。外审常问"换个人
能否复现你的结果"，控制类还要能复现稳定性与性能结论。以下方法于 2026-07-09 核验框架（见
resources/official-source-map.md）；本刊是否强制提交复现材料 待核实。
| 支柱 | 内容 |
|---|---|
| 环境可复现 | 语言/库版本、系统、硬件、随机种子固定 |
| 数据可复现 | 数据来源、划分、预处理脚本、校验和 |
| 流程可复现 | 从原始数据到论文图表的一键脚本 |
requirements.txt/environment.yml
或 Dockerfile。| 风险 | 防范 |
|---|---|
| 测试集调参 | 只在验证集调参，测试集仅最终评测一次 |
| 数据泄露 | 划分在预处理前，防未来信息进训练 |
| 重复样本跨集 | 去重，确认训练/测试无交叠 |
| 分布偏移未披露 | 说明训练与测试分布差异 |
repro/
README.md # 环境、数据、运行、预期结果、硬件
env/ # requirements.txt / environment.yml / Dockerfile
data/ # 数据或获取脚本 + 校验和
src/ # 方法与基线
configs/ # 每个实验/图表的配置与种子
scripts/ # run_all.sh：原始数据→论文图表
results/ # 关键指标，供复现校验
aas-submission）。[环境] 版本/硬件/种子固定？仿真器/步长/求解器记录？
[数据] 来源/划分/预处理/校验和齐全？无泄露/无跨集重复？
[随机] 多次运行均值±方差？随机来源已说明？
[流程] run_all.sh 可从数据跑到图表？
[超参] 搜索空间/选择准则/取值公布？基线同等调参？
[匿名] 复现材料无身份泄露？
[待核实] 本刊是否强制提交代码/数据：待核实
控制类结论（稳定、收敛、性能改进）依赖仿真设定，须逐项固定并披露：
| 项 | 记录内容 |
|---|---|
| 被控对象模型 | 状态方程/传递函数、阶数、参数取值 |
| 初始条件 | 状态初值、参考信号 |
| 数值积分 | 求解器类型、步长、容差 |
| 噪声/扰动 | 类型、幅值、随机种子 |
| 仿真时长 | 总时长、采样率 |
| 控制器参数 | 增益、自适应率、约束值 |
同一套参数应能复现论文中的每条响应曲线与性能指标；把这些写进 configs/ 与 README。
results/ 放关键指标基准值，复现时比对偏差在容许范围内。aas-camera-ready）再去匿名、正式公开，衔接 aas-artifact-evaluation。aas-camera-ready）把稳定版复现包正式公开，并在稿件中标注最终获取地址。《自动化学报》(Acta Automatica Sinica, AAS) 重视理论与实验双支撑，可复现性是让稳定性/性能结论
可信的基础。把复现包与 aas-experiments、aas-artifact-evaluation 联动整理，能显著降低外审对
"结果可靠性"的疑虑，也为见刊后同行复现打好基础。政策性要求未双源确认前标 待核实。