Skip to content

joshua-zyy/academic-paper-writer / skills/academic-polishing

academic-polishing

Verified

Polish academic prose, de-AI-ify text, control claim strength, or rewrite method sections for CS/AI/ML papers. Executes Prose Quality Gate, Claim Strength Audit, and de-AI pass. Use when: removing AI writing patterns from paper text, adjusting claim strength to match evidence level, rewriting method sections with proper narrative flow, improving academic writing quality, checking for overclaiming. Triggers on: 润色, polish, improve writing, 去AI, de-AI, claim strength, 改写, rewrite method, prose quality, 降级表述, remove AI patterns, academic writing polish, 学术润色, 去AI化, 降级结论, improve prose.

references/claim-strength.md

View on GitHub
# Claim 强度控制

## 强结论 (Strong)

**条件**(必须同时满足):
- 有本地可复核结果
- 评估协议没有明显方法学缺陷
- 或多篇 VERIFIED 文献支持同一背景事实或比较结论

**量化启发式**:
- 有独立测试集结果(非交叉验证内部划分)
- 有完整 baseline 比较(≥3 个强基线)
- 指标计算方式与标准定义一致
- 多随机种子运行,报告均值 ± 标准差

**典型表达**:show, demonstrate, outperform, consistently improves

## 中等结论 (Medium)

**触发降级的场景**:
- 结果来自内部验证而非独立测试
- 对比基线不全
- 只有单一来源支撑
- 本地代码与论文叙述尚未完全对齐

**量化启发式**:
- 仅有交叉验证或内部划分结果
- baseline 数量 <3 或缺少关键对比方法
- 单次运行无方差估计
- 验证集可能参与超参选择

**典型表达**:suggest, indicate, is associated with, appears to improve

## 弱结论或假设 (Weak)

**触发降级的场景**:
- 只有用户口述,没有本地或外部证据
- 指标无法复核
- 引文尚未核验
- 评估协议存在明显泄漏或偏乐观设计

**量化启发式**:
- 无本地运行结果,仅有文献报道
- 无 baseline 比较
- 数据集划分存在泄漏风险
- 指标定义与标准不符

**典型表达**:may, could, is hypothesized to, requires further validation

---

## empirical paper 的常见过度表述

以下说法需要特别警惕,出现时必须主动弱化措辞:

| 过度表述 | 应改为 |
|---------|--------|
| 把内部验证写成"泛化良好" | 明确标注 internal validation 边界 |
| 把验证集调阈值后的结果写成最终性能 | 区分 validation 和 test 结果 |
| 把存在受试者泄漏的结果写成 clinical-level 证据 | 标注数据泄漏风险 |
| 把尚未补全 baseline 的结果写成 SOTA | 降级为"与当前比较范围相比……" |
| 把解释性热图直接写成生物标志物结论 | 区分"模型观察"和"领域解释" |
| "significantly improves"(无 p 值/效应量) | "appears to improve" 或附具体数值差异 |
| "demonstrates strong generalization"(无外部测试) | "shows promising results on internal validation" |
| "outperforms all baselines"(基线不全) | "outperforms the compared baselines" |

---

## 自动降级触发词

以下词汇出现时,必须检查是否满足 Strong 条件;不满足则强制降级:

- "state-of-the-art" / "SOTA"
- "significantly outperforms"
- "demonstrates strong generalization"
- "robust to ..."
- "universally applicable"
- "clinically relevant"(无临床验证时)

---

## Claim Strength Audit 输出格式

```md
## Claim Strength Changes

| 原文 | 问题 | 修改后 | 原因 |
|------|------|--------|------|
| "Our method outperforms all baselines" | 基线不全 | "Our method appears to improve over the compared baselines" | 缺少关键对比方法 |
| "This demonstrates strong generalization" | 仅内部验证 | "On our internal validation set, ..." | 无外部测试集 |
| "The proposed module significantly improves accuracy" | 无统计检验 | "The proposed module improves accuracy by 2.3 percentage points" | 未报告 p 值或置信区间 |
```

---

## Overclaim 检查清单

以下词汇出现时必须标记并弱化,除非证据异常强且范围严格限定:

| 过度表述 | 安全替换 |
|---------|---------|
| `prove` | `show` / `suggest` |
| `conclusively` | `to our knowledge` |
| `unprecedented` | `to our knowledge, no prior work has ...` |
| `best` / `superior` | `among the strongest` / `competitive` |
| `first`(未验证) | `to our knowledge, this is the first ...` |
| `clinically relevant`(无临床验证) | `may have clinical implications` |

---

## 学术风格检查

### 冠词用法

- 首次提及单数可数名词:`a` 或 `an`
- 再次提及同一事物:`the`
- 泛指复数:通常不用冠词
- 唯一实体:通常用 `the`
- 抽象名词泛指:通常不用冠词

典型修复:

- 错:`The hypoxia induces ...`
- 对:`Hypoxia induces ...`

### 数字与单位

- 测量值使用数字
- 数值与单位之间留空格:`25 cm`、`3.2 s`
- 统计符号和数学记号保持一致
- 范围使用 en dash

### 学术语体

- 避免口语化填充词和弱评价语言
- 仅在学科和文体适合时使用 `we`
- 保持名词化有用但不过度
- 保持适当的客观语体,但不强求生硬

### 句段检查

- 每句应表达一个主要命题
- 从句必须依附于主句
- 不用单个逗号连接两个独立分句
- 每段需要一个主控观点和支撑材料
- 避免由 `although` 或 `whereas` 引起的常见句法错误

### 完整性规则

- 不得编造参考文献
- 不得篡改量化值(除非修正用户指出的明显笔误)
- 不得将关联升级为因果
- 不得暗示比研究支持范围更广的泛化能力
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 196kUpdated Claude CodeCodexVerifiedInstall