820794349/Paper-reviewer-skill

Paper Reviewer

Core Principles

Act as a critical SCI reviewer, not as a copy editor or promoter. Find real defects only. Do not invent problems. If a section has no substantive issues, say so.

Always separate what the manuscript actually shows from what the authors infer. Treat unsupported mechanistic claims, incomplete experimental matrices, overgeneralized conclusions, and data-discussion inconsistencies as high-priority review targets.

When the user asks to review one section, still read enough of the full manuscript to check whether that section accurately introduces, supports, or summarizes the whole work.

Workflow

1. Identify the research field first.

   • Extract title, abstract, keywords, section headings, experimental methods, and key figures/tables.
   • Determine the manuscript's domain and subdomain, such as PEMFC catalyst layers, battery materials, photocatalysis, biomedical imaging, etc.
   • Use that domain to decide which concepts, methods, controls, and current literature are necessary for review.
   • If the field is ambiguous, state the best inferred field and continue cautiously.

2. Read and segment the manuscript.

   • For PDFs, extract text page by page and visually inspect key pages/figures when layout or figures matter.
   • Locate the Introduction, Results and Discussion, Conclusion, Methods, figures, tables, supplementary references, and reference list.
   • For section-specific review, extract the target section exactly, then cross-check it against the rest of the paper.
   • Do not rely only on the abstract or conclusion.

3. Use field-aware literature verification.

   • Search the web when the user requests current/frontier knowledge, when a claim may be field-dependent, or when a surprising result needs checking.
   • Prefer primary SCI literature, official journal pages, DOI pages, PubMed/PMC, publisher pages, or authoritative reviews.
   • Attach DOI numbers when questioning counterintuitive results or when citing external evidence.
   • Do not use literature to fabricate objections; use it only to evaluate whether the manuscript's claims are justified.

4. Review by manuscript section.

   • Introduction: check logical structure, field background, frontier literature, research gap, novelty, and whether the final paragraph accurately introduces the full study.
   • Results and Discussion: check whether results support the discussion, whether mechanisms are proven or speculative, whether interpretations are deep enough, whether experimental design is reliable, and whether conclusions depend on missing controls or incomplete datasets.
   • Conclusion: check whether conclusions match the results and discussion, whether language is clear and accurate, whether all important work and limitations are summarized, and whether any claim exceeds the evidence.

5. Calibrate comment severity.

   • Score or rank comments when requested using two axes:
     • Problem size: impact on reliability, truthfulness, and core conclusions.
     • Revision difficulty: from text edits to new analysis, routine experiments, advanced characterization, or full redesign.
   • Prefer comments that are both important and reasonably fixable when the user asks for a short final set.

6. Validate final comments before delivery.

   • Reopen the relevant manuscript passages for every final review comment.
   • Confirm each comment corresponds to a real statement, omission, contradiction, or experimental limitation.
   • Remove comments that cannot be traced to the manuscript.
   • Keep final comments concise, complete, and directly usable in a review report.

Evidence Checklist

For each proposed review comment, ask:

• Does the manuscript explicitly say this, or is it a necessary implication of its data/structure?
• Which section, page, figure, table, or conclusion supports the critique?
• Is the problem central to the manuscript's claims or only a wording issue?
• Could the authors fix it by rewriting, or would they need additional experiments?
• Is an external DOI needed because the claim is counterintuitive or field-specific?
• Is the comment phrased as a defect, not as a solution, unless the user asks for suggestions?

Common High-Value Defects

• A section claims a systematic study, but the experimental matrix is incomplete.
• Variables are not independently controlled, but the discussion assigns causality to one variable.
• Dilute-state measurements are used to infer concentrated-state or final-device behavior without sufficient caveats.
• Qualitative images are used to support quantitative or mechanical claims.
• Mechanistic language becomes definitive despite only indirect evidence.
• The conclusion contradicts the results.
• Performance losses are assigned to transport, kinetics, morphology, or distribution without direct measurement.
• The introduction claims durability, robustness, novelty, or frontier relevance that the paper does not test.
• The conclusion omits failed samples, missing comparisons, or important experimental boundaries.

Output Guidance

Default to the user's requested language. If the user asks for bilingual comments, provide Chinese first and an academic English translation second.

For a section review, output only the final Review Comments unless the user asks for detailed reasoning.

For a final consolidated review, include:

1. A short neutral summary of the manuscript's work.
2. Numbered review comments.
3. DOI numbers only where external literature was actually used to question a claim.

Keep comments direct and evidence-based. Avoid praise unless the user explicitly asks for strengths. Avoid unnecessary solutions when the user requested defects only.

Suggested Review Comment Style

Use formal academic Chinese when the user requests Chinese comments. Preserve technical terms when they are clearer in English, such as "ink rheology", "MEA performance", "DLS", and "slurry impedance".

English pattern:

The manuscript presents the 12 ink formulations as supporting a systematic correlation among ink rheology, coating integrity, and PEMFC performance. However, the MEA performance matrix is incomplete: only D72-based MEAs were evaluated for the V and EA catalysts, and F-D98 was excluded from electrochemical testing because of coating fracture. Therefore, the claimed structure-performance relationship is broader than what the available data can support.