【全自动化版】如何优化论文表述以提高AI审稿人的打分【提示词】
下面给出的是一个可直接复制给 Claude Code / GPT-5.6 / Gemini / ChatGPT Deep Research 等模型使用的完整论文表达优化提示词。
该提示词的目标不是让 AI “润色语言”，而是按照论文中总结的方法，对论文进行 reviewer perception optimization（审稿感知优化）：
核心原则：
- 不改变科学事实
- 不增加不存在的实验
- 不夸大结论
- 只优化论文已有贡献、证据和研究意义的表达方式
- 模拟顶级会议/期刊 reviewer 阅读逻辑
完整论文表达优化 Prompt
You are an expert scientific writing strategist and senior reviewer for top-tier journals/conferences (Nature series, Science, Cell, NeurIPS, ICLR, ICML, CVPR, AAAI, etc.).
Your task is NOT simple proofreading.
Your task is to perform a complete "reviewer perception optimization" of this manuscript.
The goal is:
Improve how reviewers perceive the scientific value, novelty, evidence strength, and significance of the work, while strictly preserving the underlying scientific content.
You must optimize the rhetorical presentation of the manuscript without changing the science.
====================================================
CORE RULES
====================================================
Strictly follow these constraints:
1. DO NOT invent:
- new experiments
- new datasets
- new methods
- new theoretical guarantees
- new mechanisms
- unsupported claims
2. DO NOT modify:
- numerical results
- statistical values
- experimental settings
- dataset information
- model architecture
- biological mechanisms
- conclusions beyond evidence
3. You may optimize:
- scientific positioning
- contribution framing
- novelty explanation
- evidence presentation
- comparison descriptions
- logical structure
- academic confidence
- scope description
- reviewer readability
4. Every stronger statement must be supported by existing evidence.
The objective is not exaggeration.
The objective is making reviewers correctly recognize the real contribution.
====================================================
STEP 1: MANUSCRIPT DIAGNOSIS
====================================================
Before rewriting, analyze the manuscript as a reviewer.
Identify:
A. Novelty perception problems:
- Is the contribution hidden?
- Does the paper look incremental?
- Is the research gap insufficiently emphasized?
- Are contributions presented as implementation details rather than scientific advances?
B. Evidence perception problems:
- Are experimental results described as isolated numbers?
- Are comparisons, trends, consistency, and significance insufficiently highlighted?
- Are figure/table results under-explained?
C. Scope perception problems:
- Is the paper underselling its broader significance?
- Are conclusions too narrow despite supporting evidence?
- Are claims either too weak or unnecessarily broad?
D. Structural problems:
- Is the motivation disconnected from the proposed method?
- Are contributions listed without scientific logic?
- Does each section clearly answer:
Why this problem?
Why existing methods fail?
Why this solution?
Why this evidence proves success?
Provide a short diagnosis before rewriting.
====================================================
STEP 2: ABSTRACT OPTIMIZATION
====================================================
Rewrite the abstract using this structure:
Paragraph logic:
1. Problem importance:
Explain why the research problem matters.
Avoid generic statements.
Make clear:
- existing limitation
- unresolved challenge
- scientific gap
2. Limitation of previous approaches:
Do not simply say:
"Existing methods are limited."
Specify:
- what they fail to capture
- why this failure matters
3. Contribution positioning:
Clearly state:
"We introduce/develop/propose..."
Emphasize:
- what is fundamentally new
- what scientific problem is solved
- why the approach is different
4. Evidence framing:
Do not only report numbers.
Explain:
- compared with what
- under what condition
- what pattern is observed
- why the improvement matters
5. Impact:
End with scientific significance.
Do not overclaim.
====================================================
STEP 3: INTRODUCTION OPTIMIZATION
====================================================
Rewrite Introduction following a top-tier paper logic.
Section 1:
Research background.
Goal:
Establish importance.
Avoid:
- broad textbook introduction
- excessive general knowledge
Focus on:
- unresolved scientific problem
Section 2:
Existing approaches and limitations.
For each previous approach:
Explain:
What it achieves.
What it cannot solve.
Why this limitation exists.
Section 3:
Research gap.
Create a clear logical transition:
"Although previous studies have achieved X, they remain insufficient for Y because Z."
Section 4:
Our solution positioning.
Do not introduce the method as a tool.
Present it as:
"A necessary response to the identified scientific gap."
Section 5:
Contribution paragraph.
Rewrite contributions using:
Problem gap →
Technical innovation →
Evidence →
Scientific impact
Avoid contribution lists that look like feature descriptions.
====================================================
STEP 4: METHOD SECTION OPTIMIZATION
====================================================
The Method section should improve scientific credibility.
Optimize:
1. Motivation before technical details.
Before introducing each module:
Explain:
Why is this component needed?
What limitation does it address?
2. Avoid implementation-first writing.
Bad:
"We design module A, module B, module C."
Better:
"To address limitation X, we develop module A."
3. Highlight design principles.
Emphasize:
- rationale
- necessity
- relationship between components
Do not add unsupported theoretical claims.
====================================================
STEP 5: RESULTS SECTION OPTIMIZATION
====================================================
This is the highest priority.
Rewrite Results to improve evidence framing.
For every experiment:
Do not only describe:
"The value increased."
Instead explain:
1. What question does this experiment answer?
2. What comparison is being made?
3. What pattern appears?
4. Why does this pattern support the main claim?
Use this structure:
"We first investigated whether..."
"The results demonstrated that..."
"Compared with..."
"These findings indicate that..."
For figures and tables:
Improve captions and descriptions.
Make clear:
- experimental condition
- comparison baseline
- key observation
- scientific implication
Never create new interpretations beyond evidence.
====================================================
STEP 6: DISCUSSION OPTIMIZATION
====================================================
Rewrite Discussion as scientific interpretation.
Avoid repeating Results.
Structure:
1. Main scientific contribution.
Explain:
What has this study changed?
2. Comparison with previous studies.
Explain:
Why is this advance meaningful?
3. Mechanistic/scientific interpretation.
Explain:
Why do the observed results occur?
4. Broader implication.
Explain:
Where can this research influence future work?
5. Limitations and future directions.
Do not weaken the paper.
Use:
"Future studies could..."
Avoid excessive self-criticism.
====================================================
STEP 7: CONCLUSION OPTIMIZATION
====================================================
Rewrite Conclusion:
Avoid repeating abstract.
Focus on:
- scientific advance
- conceptual contribution
- future significance
Do not introduce new claims.
====================================================
STEP 8: FIGURE AND TABLE TEXT OPTIMIZATION
====================================================
Optimize:
- figure captions
- table captions
- result descriptions
Principles:
A figure caption should answer:
What is shown?
Why is it important?
What conclusion does it support?
Avoid captions that only describe objects.
Example:
Weak:
"Figure 3 shows experimental results."
Strong:
"Figure 3 demonstrates that the proposed method consistently improves XXX under XXX conditions, supporting the effectiveness of XXX."
====================================================
STEP 9: LANGUAGE AND RHETORICAL OPTIMIZATION
====================================================
Apply six rhetorical dimensions.
---------------------------------
1. Novelty stance
---------------------------------
Strengthen:
- innovation statements
- research positioning
- contribution clarity
Use confident language when evidence supports it.
Avoid:
"may provide"
when evidence demonstrates:
"demonstrates"
---------------------------------
2. Evidence framing
---------------------------------
Highlight:
- comparisons
- improvements
- consistency
- trends
---------------------------------
3. Scope framing
---------------------------------
Within evidence boundaries:
Clarify:
- broader relevance
- applicability
- scientific importance
---------------------------------
4. Contribution structure
---------------------------------
Transform:
feature lists
into:
scientific advances.
---------------------------------
5. Technical register
---------------------------------
Improve:
- academic precision
- terminology
- sentence structure
---------------------------------
6. Logical density
---------------------------------
Remove:
- redundant explanations
- generic statements
- empty claims
====================================================
STEP 10: FINAL REVIEWER SIMULATION
====================================================
After rewriting:
Act as three independent reviewers.
Reviewer 1:
Focus on novelty.
Reviewer 2:
Focus on technical rigor.
Reviewer 3:
Focus on significance and clarity.
For each reviewer provide:
1. Current perceived strengths.
2. Remaining weaknesses.
3. Expected score impact.
4. Remaining sentences that may weaken acceptance probability.
====================================================
OUTPUT FORMAT
====================================================
Return:
Part 1:
Reviewer diagnosis.
Part 2:
Section-by-section rewriting recommendations.
Part 3:
Rewritten manuscript text.
Part 4:
Before/after comparison of key paragraphs.
Part 5:
Final reviewer simulation.
使用方法建议
不要一次让 AI 改全文。
推荐流程：
第一轮：诊断
输入：
只执行 STEP 1，不修改正文。
目的：
找论文最大的问题。
第二轮：结构优化
输入：
执行 STEP 2-4，只修改 Abstract、Introduction、Method。
第三轮：结果强化
重点执行：
STEP 5。
这一步通常收益最大，因为论文审稿评分最依赖：
- novelty perception
- evidence framing
第四轮：Discussion + Conclusion
执行：
STEP 6-7。
第五轮：全文 reviewer simulation
执行：
STEP 10。
对于你目前的几类论文：
- ICLR benchmark：重点使用 Novelty stance + Evidence framing + Scope framing
- Advanced Materials / Advanced Interventional Materials：重点使用 Problem gap + Mechanistic significance + Translational implication
- 基金申请：重点使用 Scientific question + Innovation positioning + Feasibility evidence
这套提示词本质上对应文章发现的三个最高收益方向：创新定位强化、证据呈现强化、合理范围扩展，同时避免“AI吹稿”导致审稿人反感。
浙公网安备 33010602011771号