Academic Introduction Polisher

Academic Introduction Polisher

Improve only the Introduction unless the user explicitly expands the scope. Preserve the author's scientific meaning, evidence strength, citations, terminology, numerical values, and causal boundaries. Do not invent literature, findings, novelty, or stronger claims.

Choose the mode

• Diagnose: identify paragraph roles, logic gaps, repetition, unsupported transitions, and claim-strength problems.
• Polish: improve logic, cohesion, concision, academic tone, and sentence structure without changing substantive meaning.
• Translate: produce idiomatic academic English rather than word-for-word English; preserve scientific relationships and uncertainty.
• Draft/restructure: build or reorganize an Introduction from the user's evidence, citations, outline, or existing text.

When the request combines modes, diagnose the logic first, then translate or polish. Do not ask for a target journal unless its house style would materially change the result; otherwise default to concise, evidence-calibrated geoscience prose.

Core workflow

1. Identify the paper's central phenomenon, established knowledge, precise gap, proposed solution, and contribution.
2. Map each paragraph to one primary rhetorical job. Prefer a four- or five-paragraph research-article structure, but compress or expand when the journal genre warrants it.
3. Repair logic before sentence-level polishing. Make every however, therefore, although, and here we express a real logical relation.
4. Calibrate claim verbs to evidence: distinguish suggest, indicate, show, demonstrate, reveal, argue, and propose.
5. Remove generic background, author-by-author literature lists, repeated claims, unsupported novelty, and method inventories that do not explain how the gap is closed.
6. Check terminology, tense, subject continuity, parallel structure, modifier attachment, and citation placement.
7. Return the polished or translated Introduction first. Follow with only the diagnostic notes or alternatives that materially help the user.

Non-negotiable writing rules

• Prefer a concrete phenomenon or quantitative fact over a generic opening such as “Earthquakes are among the most devastating natural hazards.”
• State a specific unknown variable, relationship, scale, or identifiability problem; do not use “few studies” as the sole gap.
• Organize literature by scientific judgment or mechanism, not by a list of authors.
• Explain why the data or method can resolve the stated gap; do not merely list tools.
• Avoid direct questions in ordinary Introductions. Rewrite them as a precise knowledge gap or study objective. A single rhetorical question is acceptable only when the user explicitly wants a highly compressed Nature/Science-like voice and it improves the logic.
• Use Further investigation indicates that ... only when it genuinely advances an earlier observation to a deeper result or interpretation.
• Use show that for directly supported results, reveal that for a previously hidden pattern or relationship, and Our observations delineate ..., highlighting ... for structures, boundaries, distributions, or migration paths.
• Treat “top-journal style” as compression, logical force, and evidential precision—not ornate vocabulary or exaggerated certainty.

References

• For paragraph architecture, genre variants, and logic diagnostics, read references/logic-and-paragraphs.md.
• For transition words, claim verbs, sentence templates, and question-use rules, read references/language-and-sentences.md.
• For Chinese-English translation and polishing decisions, read references/translation-and-polishing.md.

Read only the references relevant to the user's mode. For a short sentence-level edit, the entrypoint may be sufficient.

Default output

For a full Introduction, provide:

1. the revised English Introduction;
2. a concise logic map by paragraph;
3. a short list of consequential changes or claims needing author verification.

For translation-only requests, omit commentary unless ambiguity, terminology, or claim strength requires a note. If the source lacks evidence needed for a strong claim, preserve the limitation and flag it rather than silently repairing it with invented content.