审阅修改后的学术稿件
逐项核对审稿意见、作者回复与修订文本,检查新增问题和内容一致性,给出发表建议及剩余修改要求。
完整提示词
<context>
A second-round review has a narrower job than the first: decide whether the authors dealt with the issues raised, not re-review the paper from scratch or raise points that could have been raised the first time. Good second-round reviewers check the response letter against the actual text, because responses sometimes describe changes that never made it into the manuscript, soften a claim in the letter but not in the abstract, or add an analysis that answers a different question. Revisions also introduce new problems: a new analysis with its own flaws, numbers that no longer match between abstract, tables and text, or a reframed contribution. A reasonable rebuttal deserves acceptance even if you would have done it differently. Editors dislike reviewers who move the goalposts.
</context>
<task>
Review this revision.
<previous_review>
{{my_previous_review}}
</previous_review>
<response_letter>
{{response_letter}}
</response_letter>
<revised_text>
{{revised_text}}
</revised_text>
{{#other_reviews}}
<other_reviews>
{{other_reviews}}
</other_reviews>
{{/other_reviews}}
1. List every comment from my previous review, numbered as in the original.
2. For each, find the authors' response and the corresponding change in the revised text. Classify it: resolved, partly resolved, rebutted reasonably, rebutted unconvincingly, or not addressed.
3. Flag any response that claims a change you cannot find in the revised text, quoting the claim and naming where you looked.
4. Check consistency after revision: abstract, results, tables, figures and conclusions still agree, and softened claims are softened everywhere.
5. Review only the new or changed material for new problems, for example a new analysis, new data or a reframed contribution.
6. Decide what still blocks publication and recommend accept, minor revision, major revision or reject.
</task>
<constraints>
- Do not raise new issues about unchanged material unless they are serious enough to change the conclusions; if you do, label them clearly as new and explain why they were not raised before.
- Accept a reasonable rebuttal even if you disagree on taste; reserve "unconvincing" for rebuttals that leave a conclusion unsupported.
- Quote the response and the revised text when judging; never assume a change was made or not made without looking.
- If only part of the revised manuscript was provided, say which comments cannot be checked.
- Start with one line reminding the reviewer that the manuscript is confidential and to check the venue's AI-use policy.
</constraints>
<output_format>
One reminder line, then:
## Status of previous comments
A table: comment | response summary | where in the text | status | note.
## Claims in the response not found in the text
Bullets with quotes, or "None found".
## New issues from the revision
Numbered: location, problem, why it matters, what would resolve it.
## Remaining major issues
Numbered, or "None".
## Recommendation
The recommendation and the deciding reasons in two or three sentences.
</output_format>可替换内容
- {{my_previous_review}}
- {{response_letter}}
- {{revised_text}}
- {{#other_reviews}}
- {{other_reviews}}
- {{/other_reviews}}
分类与标签
学习教育、学术审稿、论文修改、同行评议、发表建议