Community写作与编辑github.com

academic-paper-writing

指导英文 SCI/SSCI 学术论文的构思、写作与投稿全流程。适用于撰写或修改英文期刊论文(题目、摘要、引言、文献综述、方法、结果讨论、结论、参考文献)、论文逻辑与语言润色、选刊投稿、回复审稿意见、拒稿改投等场景。源自资深教授(交通/规划领域,45+ 期刊审稿人)的论文写作与投稿课程方法论。

academic-paper-writing 是什么?

academic-paper-writing is a Claude Code agent skill that 指导英文 SCI/SSCI 学术论文的构思、写作与投稿全流程。适用于撰写或修改英文期刊论文(题目、摘要、引言、文献综述、方法、结果讨论、结论、参考文献)、论文逻辑与语言润色、选刊投稿、回复审稿意见、拒稿改投等场景。源自资深教授(交通/规划领域,45+ 期刊审稿人)的论文写作与投稿课程方法论。.

兼容平台~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/LDLZM/academic-paper-writing/tree/main

Installed? Explore more 写作与编辑 skills: steipete/notion, langchain-ai/langchain, bytedance/podcast-generation · View all 6 →

在你喜欢的 AI 中提问

打开一个已预加载此 Agent Skill 的新对话。

文档

Academic Paper Writing — English Paper Writing & Submission

Core Principles (Throughout the Entire Workflow)

  • Tell a Story: A paper is essentially storytelling, with the three phases—conception, writing, submission—tightly interlinked.
  • Be a Lawyer/Debater: Every paragraph is a "conviction"—with evidence, facts, and logical progression; sentences must connect without leaps.
  • Reviewer's Perspective: Submission is a process of "pleasing the reviewer." Reviewers look at "appearance" (structure, language, logic, figures/tables) and "substance" (innovation, contribution, added value). Seeing "interesting" in reviewer comments is the highest praise.
  • Input Before Output: Read at least 50 papers in your target field before writing; maintain daily reading to keep language sense. No input, no output.

Three Phases · Ten Elements Overview

Phase 1: Conception

  1. Novelty (Novelty / Interesting): Not necessarily groundbreaking innovation, but must have novelty—new application, new method, new data, new perspective—at least one.
  2. Three Key Positioning Words—Innovation / Contribution / Added Value:
    • Innovation is hard, contribution is easier: even without a new method, you can make a "contribution" with new data, old problem + new conclusion, or a new perspective;
    • Having innovation and contribution still requires added value: the method must genuinely be better to use, the case must truly deepen understanding of the scientific problem;
    • At least one of these three must be present, otherwise it's "publishing for the sake of publishing." Before writing, answer: "What is the significance of this paper?"
  3. Idea Sources: Listen more, read more, ask more—academic talks, editor-in-chief meetings, peer discussions; summarize controversial topics, compare differences in conclusions from different methods/data/regions. Once an idea emerges, immediately solidify it on paper: what data to use, what results to expect, to prevent forgetting the next day.

Phase 2: Writing

Classic five sections: Title → Abstract → Introduction (with/without Literature Review) → Research Design (Data & Methods) → Results & Discussion (including Policy Implications) → Conclusion.

Detailed writing methods, sentence templates, and common pitfalls for each section are in references/writing-guide.md:

  • Title: combination of elements + -ing openers + colon progression; avoid "study on/research on"; avoid self-praise with "novel/innovative"; question format allowed.
  • Abstract: five-part structured (background 3 sentences → method & data → results → conclusion → significance & contribution); contribution must be specific, avoid vague generalities.
  • Introduction: four-paragraph structure (background paragraph → Gap paragraph → Task paragraph → Contribution paragraph); the first sentence of the Gap paragraph must be a negative statement that hits the mark ("remains unclear / has not / remains limited").
  • Literature Review: sub-sections corresponding one-to-one with gaps + a concluding summary section; strictly avoid "what Zhang did" chronicles; build progressive links and inductive connections.
  • Methods: English papers emphasize reproducibility; clarity ≠ simplicity.
  • Results & Discussion: core term is "interaction"—compare, confirm, refute previous work, and provide Policy Implications.
  • Conclusion: three-plus-one (conclusion paragraph + contribution paragraph + limitations + future study), echoing the introduction; limitations and future work written separately; contribution should aim high (shift/advance/promote), but avoid exaggeration like "first, fills a gap, innovative."
  • Language & Logic: write directly in English, strictly avoid Chinese-to-English translation; prefer short sentences; each sentence should interlock.

Phase 3: Submission

Detailed operations for journal selection, review process, responding to comments, and handling rejection are in references/submission-guide.md:

  • Indexing Systems: Web of Science (SCI/SSCI/A&HCI), Scopus, EI; publishing groups (Elsevier, Springer, Taylor & Francis, Sage) produce data, indexing institutions transport data.
  • Journal Selection: First check the school/graduation catalog → similar papers → Web of Science search → Journal Finder (third-party tools are more comprehensive than publishers' own recommendations); weigh subject fit (most important), review timeline, impact factor, acceptance rate.
  • Review Process: Editorial office initial check (plagiarism/format/language) → Editor-in-Chief/Associate Editor assesses novelty and decides to send for review or reject → Invite 2–5 reviewers (review period generally 21–30 days) → Editor-in-Chief synthesizes comments to make a decision. Single round typically takes 2–6 months, time primarily spent finding reviewers and author revisions.
  • Responding to Reviewer Comments: Three categories for handling: (1) agree and feasible (thank + revise + indicate page and line numbers); (2) agree but not feasible (acknowledge + explain minimal impact on conclusions or defer to future research); (3) disagree (polite, present facts and reasoning, cite literature, avoid confrontation). Use "Thank you for your comments" and second-person "you." Final decision rests with the Editor-in-Chief.
  • Rejection: Mainstream path is to resubmit to another journal (reconstruct the story according to the new journal, especially title and abstract); in rare cases (most reviewers positive, fatal points not severe) you may gather courage to resubmit with a full response letter.

References & Tools

  • Citation Strategy Mix Old and New: In the first two paragraphs of the introduction, avoid overly old citations (affects timeliness impression); "old" must be at the "bible level" for that field (foundational for concepts/methods), otherwise cite recent relevant applications (e.g., K-means need not cite the original paper).
  • Cite 2–3 papers from the target journal: give credit to the editor-in-chief and help find reviewers; but don't overdo it to avoid raising self-citation rates.
  • Reference Management: Preferred: EndNote (Word plugin, cloud sync, team sharing, extensive interfaces); Alternatives: Mendeley, Zotero; platforms can be inter-converted.
  • Figures & Tables: Figures should be self-contained with complete annotations; one figure is worth a thousand words; use three-line tables; tool combinations: Python / MATLAB / R / SPSS (export statistical tables directly) / Excel / Visio / draw.io.
  • Formatting: Follow the journal template; page numbers and line numbers are mandatory; common requirements: Times New Roman, ≥12 pt font, double spacing.
  • AI Assistance: Can be used to generate alternative titles and organize literature, but prompts must include structural constraints (no chronicling, sentence interlocking, progressive layering); do not rely on Chinese-to-English translation; maintain hands-on writing practice.

Usage Instructions

Route according to the user's current stage:

User RequirementAction
Topic selection / idea generation / evaluating idea valueUse Phase 1's "Novelty + Three Keywords" framework to evaluate, give interesting rating and improvement suggestions
Draft title / write abstract / write introduction / write literature review / write conclusion / polish a sectionRead corresponding section in references/writing-guide.md, apply structure templates and sentence patterns, check each pitfall
Full paper diagnosis / logic reviewFollow the "lawyer conviction" standard to examine each paragraph: check if sentences interlock, if the Gap paragraph uses negative statement directly, if contributions are overly vague
Journal selection / submission / responding to reviewer comments / handling rejectionRead references/submission-guide.md, follow the corresponding workflow and phrasing templates

Always diagnose before writing: before revising any paragraph, first point out which structural/logical rule it violates, then provide a revised version and reasoning.

相关技能