全部数字来自 502 篇获奖论文（2004–2024）与 3 篇官方评委点评的统计口径。
[背景 1 句] As [现象/趋势] [变化方向], [研究对象] faces [具体挑战]. [目标 1 句] To explore/evaluate/predict [核心问题], we develop [模型名] model. [数据 1–2 句] Before modeling, we [数据预处理动作：classified N species into M categories / estimated parameters from X sources]. [Task 1] For Task 1, we established [模型名] based on [基础理论], incorporating [关键创新点]. The results show that [量化结论，含数字]. [Task 2] For Task 2, we calculated [指标1], [指标2] and [指标3] to evaluate [维度1–3]. Results indicate that [趋势/数值结论]. [Task 3] For Task 3, we compared [N] scenarios and found that [具体结论，含数字]. [Task 4/N] We varied [参数] from [A] to [B]; the results show [具体趋势/数值]. [灵敏度 1 句] Sensitivity analysis on [参数名] confirms the model is [stable/robust]. [收尾 1 句] We also discuss [优势] and acknowledge [主要局限], concluding that [总体结论/建议]. Keywords: [模型名], [核心概念1], [核心概念2], [核心概念3]
| 指标 | 中位数 | O 奖推荐 |
|---|---|---|
| 总词数 | 437 | 400–500 |
| 具体数字个数 | 20–30 | 20–35 |
| 含数字的句子 | 6–8 | ≥ 8 |
| 模型名出现 | 2–4 次 | ≥ 3 次 |
| 关键词 | 4–5 | 4–5，含模型名 |
| Task 覆盖 | 100% | 100% + 各 ≥ 1 数字 |
| 页 | 内容 |
|---|---|
| 1 | Summary（1 页，不加图） |
| 2 | Contents |
| 3–4 | Introduction（2 页为限） |
| 5 | Assumptions + Notations |
| 6–8 | Task 1 建模 2 页 + 结果 1 页 |
| 9–11 | Task 2 建模 + 验证 |
| 12–15 | Task 3（含指标定义） |
| 16–17 | Discussion / 关键结论页 |
| 18–20 | Task 4 |
| 21–22 | Task 5 / 6 |
| 23 | 灵敏度分析 |
| 24 | Strengths & Weaknesses |
| 25 | References |
收尾必须有一句结论：the model is stable / most sensitive to X。只画图不写结论等于没做。