academic-research

# Cross-Disciplinary Academic Research Skill Pack / 跨学科学术研究技能包 Cross-disciplinary end-to-end academic research pipeline: from idea evaluation to camera-ready submission, covering all disciplines. 跨学科端到端学术研究流水线：从构思评估到终稿提交，覆盖全学科。 ## When to Activate / 触发条件 - User mentions writing a paper, thesis, or academic document / 用户提到写论文、学位论文或学术文档 - User asks about LaTeX formatting, compilation, or journal templates / 用户询问 LaTeX 排版、编译或期刊模板 - User requests literature review or citation management / 用户请求文献综述或引文管理 - User triggers `/academic` command / 用户触发 `/academic` 命令 - User mentions keywords: paper, thesis, latex, bibtex, conference, journal, arxiv, databases, integrity, benchmark, reproduce, vibe, causal, 论文, 学术, 文献综述, 期刊, 开题, rebuttal, 复现, 基准, 研究诚信, 数据库, AI协作 ## Architecture Overview / 架构概览 ``` ┌───────────────────────────────────────────────────────┐ │ /academic Command (22 sub-cmds) │ │ /academic 命令 (22 个子命令) │ ├───────────────────────────────────────────────────────┤ │ │ │ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ │ │ │ Idea │ │ Lit │ │ Paper │ │ Experiment│ │ │ │ Eval │ │ Review │ │ Writing │ │ Design │ │ │ │ 构思评估 │ │ 文献综述 │ │ 论文写作 │ │ 实验设计 │ │ │ └────┬─────┘ └────┬─────┘ └────┬─────┘ └────┬─────┘ │ │ │ │ │ │ │ │ ┌────┴─────────────┴────────────┴─────────────┴────┐ │ │ │ Scientific Database Layer / 科学数据库层 │ │ │ │ 28+ databases via MCP | arXiv PubMed OpenAlex │ │ │ │ 28+ 数据库 MCP 集成 | FRED ADS SSRN ... │ │ │ └────────────────────┬─────────────────────────────┘ │ │ │ │ │ ┌────────────────────┴─────────────────────────────┐ │ │ │ Domain Methods / 领域方法层 │ │ │ │ Causal Inference | De-AIGC | Vibe Research │ │ │ │ 因果推断 | 降AIGC | AI协作 │ │ │ └────────────────────┬─────────────────────────────┘ │ │ │ │ │ ┌────────────────────┴─────────────────────────────┐ │ │ │ LaTeX Compilation Engine / 编译引擎 │ │ │ │ auto-detect TeXLive / Overleaf / Docker │ │ │ │ 自动检测 TeXLive / Overleaf / Docker │ │ │ └────────────────────┬─────────────────────────────┘ │ │ │ │ │ ┌────────────────────┴─────────────────────────────┐ │ │ │ Journal Template Layer / 期刊模板层 │ │ │ │ NeurIPS|ICML|ICLR|ACL|AAAI|IEEE|ACM|... │ │ │ │ 12+ venues across all disciplines │ │ │ └────────────────────┬─────────────────────────────┘ │ │ │ │ │ ┌────────────────────┴─────────────────────────────┐ │ │ │ Quality Gate System / 质量关卡系统 │ │ │ │ 5-dim pre-submit | 7-dim review | integrity │ │ │ │ 五维投前审查 | 七维评审 | 研究诚信 │ │ │ └───────────────────────────────────────────────────┘ │ │ │ │ ┌───────────────────────────────────────────────────┐ │ │ │ Cross-Discipline Coverage / 跨学科覆盖 │ │ │ │ CS|Medicine|Physics|Social Science|Economics|... │ │ │ │ 计算机|医学|物理|社会科学|经济|工程|人文|... │ │ │ └───────────────────────────────────────────────────┘ │ └───────────────────────────────────────────────────────┘ ``` ## Workflow Sub-Commands / 子命令工作流 ### 1. `check-env` — Environment Detection / 环境检测 Auto-detect LaTeX installation and available packages. 自动检测 LaTeX 安装和可用包。 ```bash # macOS / Linux bash scripts/check-environment.sh # Windows PowerShell powershell scripts/check-environment.ps1 ``` Outputs / 输出: compilers (pdflatex/xelatex/lualatex/latexmk), bibliography tools (bibtex/biber), key packages, conference style files / 编译器、参考文献工具、核心包、会议样式文件。 **Script / 脚本**: `scripts/check-environment.sh` (macOS/Linux) / `scripts/check-environment.ps1` (Windows) **Reference / 参考**: `references/latex-environment.md` ### 2. `lit-review` — Literature Review / 文献综述 **Trigger / 触发**: `/academic lit-review --topic "..."` **Workflow / 工作流** (see `workflows/literature-review.md`): 1. Seed paper identification via arXiv API + Semantic Scholar / 通过 arXiv API + Semantic Scholar 确定种子论文 2. Iterative breadth-then-depth search (forward/backward citation traversal) / 迭代广度-深度搜索（前向/后向引文遍历） 3. Abstract screening and relevance scoring (0-10) / 摘要筛选和相关性评分 4. Full-text analysis of top-ranked papers / 高分论文全文分析 5. Synthesis into structured review with BibTeX entries / 综合为结构化综述并生成 BibTeX 6. Gap identification and research positioning / 研究空白识别与研究定位 **Tools / 工具**: - `scripts/search_arxiv.py` — arXiv search with BibTeX output / arXiv 搜索并输出 BibTeX - Semantic Scholar API (free, no key needed / 免费，无需密钥) - DOI resolution via doi.org / 通过 doi.org 解析 DOI **Reference / 参考**: `workflows/literature-review.md`, `references/scientific-databases-guide.md` ### 3. `paper` — Paper Writing / 论文写作 **Trigger / 触发**: `/academic paper --venue <venue>` **Workflow / 工作流** (see `workflows/paper-writing.md`): 1. Select template from `templates/<venue>/` / 从模板目录选择目标会议模板 2. Generate paper structure / 生成论文结构（abstract → intro → method → experiments → conclusion） 3. Section-by-section writing with quality checkpoints / 逐节写作，带质量检查点 4. Insert figures, tables, algorithms via TikZ/algorithm2e / 插入图表、算法 5. Compile with auto-detected engine / 使用自动检测的引擎编译 6. Self-review with 7-dimension reviewer simulation / 7 维度审稿人模拟自审 7. Iterate until quality gate passes / 迭代直到通过质量关卡 **Writing Principles / 写作原则** (see `references/writing-guide.md`): - Gopen & Swan 7 principles of reader expectations / Gopen & Swan 读者期望七原则 - Perez micro-tips for clarity / Perez 清晰写作微技巧 - Lipton word choice guidelines / Lipton 用词指南 - One idea per paragraph, topic sentence first / 每段一个观点，主题句在前 **Reference / 参考**: `workflows/paper-writing.md`, `references/writing-guide.md` ### 4. `experiment` — Experiment Design / 实验设计 **Trigger / 触发**: `/academic experiment` **Workflow / 工作流** (see `workflows/experiment-design.md`): 1. Claims-to-experiments mapping / 论点到实验的映射 2. Baseline selection and justification / 基线选择与论证 3. Evaluation protocol design (metrics, datasets, splits) / 评估协议设计 4. Statistical significance testing plan / 统计显著性检验计划 5. Ablation study design / 消融实验设计 6. Experiment script scaffolding / 实验脚本脚手架 **Reference / 参考**: `workflows/experiment-design.md`, `references/hypothesis-generation-guide.md` ### 5. `rebuttal` — Rebuttal Writing / Rebuttal 写作 **Trigger / 触发**: `/academic rebuttal` **Workflow / 工作流** (see `workflows/rebuttal.md`): 1. Parse reviewer comments into structured items / 将审稿意见解析为结构化条目 2. Classify: accept/argue/clarify/experiment / 分类：接受/辩论/澄清/实验 3. Draft point-by-point response / 起草逐条回复 4. Plan additional experiments if needed / 如需要则规划额外实验 5. Generate camera-ready diff / 生成终稿差异 **Reference / 参考**: `workflows/rebuttal.md` ### 6. `compile` — LaTeX Compilation / LaTeX 编译 **Trigger / 触发**: `/academic compile <file.tex>` **Engine Selection Logic / 引擎选择逻辑**: ``` if document uses ctex/xeCJK → xelatex # 含中文内容 else if document uses fontspec → lualatex # 自定义字体 else → pdflatex # 标准英文论文 ``` **Compilation Pipeline / 编译流水线**: ```bash # macOS / Linux bash scripts/compile.sh -e <engine> -b -c <file.tex> # Windows PowerShell powershell scripts/compile.ps1 -TexFile <file.tex> -Engine <engine> -Bib -Clean ``` Auto-runs / 自动执行: multiple passes + bibliography + cross-reference resolution + auxiliary cleanup / 多遍编译 + 参考文献 + 交叉引用 + 辅助文件清理。 **Script / 脚本**: `scripts/compile.sh` (macOS/Linux) / `scripts/compile.ps1` (Windows) **Reference / 参考**: `references/latex-environment.md` ### 7. `idea-eval` — Idea Evaluation / 构思评估 **Trigger / 触发**: `/academic idea-eval` Evaluate research ideas using a four-layer audit (fatal flaws → lifecycle matching → five-dimension scoring → paradigm shift detection). 使用四层审计评估研究构思（致命缺陷 → 生命周期匹配 → 五维打分 → 范式跃迁探测）。 **Output / 输出**: PROCEED / PROCEED WITH GUARDRAILS / REJECT AND PIVOT verdict with evidence. 产出：带有证据的通过/带条件通过/拒绝并转向 结论。 **Reference / 参考**: `references/idea-evaluation.md` ### 8. `figure-design` — Figure Design Advisor / 图表设计顾问 **Trigger / 触发**: `/academic figure-design` Design guidance for three load-bearing figures: Motivated Example, Solution Overview, Experimental Results. 三张承重图设计指导：动机示例图、方案总览图、实验结果图。 **Output / 输出**: Per-figure design principles, anti-patterns, tool recommendations, TikZ templates. 产出：每张图的设计原则、反模式、工具推荐、TikZ 模板。 **Reference / 参考**: `references/figure-design-guide.md` ### 9. `de-aigc` — De-AIGC Review / 降 AIGC 审查 **Trigger / 触发**: `/academic de-aigc` Five-dimension assessment and rewrite guidance to reduce AI detection rates in academic writing (Chinese and English). 五维度评估和改写指导，降低学术论文 AI 检测率（中英文）。 **Output / 输出**: Pattern detection report, five-dimension scores, section-specific rewrite suggestions. 产出：模式检测报告、五维度评分、分章节改写建议。 **Reference / 参考**: `references/de-aigc-guide.md` ### 10. `causal` — Causal Inference / 因果推断 **Trigger / 触发**: `/academic causal` Causal inference method selection and analysis pipeline (DID/IV/RDD/PSM/SCM/DML). Integrates with StatsPAI (900+ functions). 因果推断方法选择和分析流水线（DID/IV/RDD/PSM/SCM/DML）。集 StatsPAI（900+ 函数）。 **Output / 输出**: Method recommendation, diagnostic tests checklist, robustness battery, publication-ready output. 产出：方法推荐、诊断检验清单、稳健性检验组合、发表级输出。 **Reference / 参考**: `references/causal-inference-guide.md` ### 11. `template` — Template Management / 模板管理 **Trigger / 触发**: `/academic template <venue>` Copies the appropriate template to the working directory / 将对应模板复制到工作目录。 **Available Templates / 可用模板**: | Venue 会议/期刊 | Type 类型 | Style 样式 | Page Limit 页数限制 | |-------|------|-------|------------| | NeurIPS | Conference 顶会 | neurips.sty | 9 + appendix 附录 | | ICML | Conference 顶会 | icml2026.sty | 8 + appendix 附录 | | ICLR | Conference 顶会 | iclr2026_conference.sty | 9 + appendix 附录 | | ACL/EMNLP | Conference 顶会 | acl.sty | 8 + appendix 附录 | | AAAI | Conference 顶会 | aaai2026.sty | 8 | | IEEE (Trans/Conf) | Journal/Conf 期刊/会议 | IEEEtran.cls | varies 不限 | | Springer (LNCS) | Conference 会议 | llncs.cls | 12-15 | | Elsevier | Journal 期刊 | elsarticle.cls | varies 不限 | | ACM | Conference 顶会 | acmart.cls | 10-12 | | APS (Phys Rev) | Journal 期刊 | revtex4-2.cls | varies 不限 | | Chinese Thesis 中文学位论文 | Thesis 学位论文 | ctexbook | varies 不限 | | Beamer Slides 学术幻灯片 | Presentation 演示 | beamer.cls | slides 页数 | **Reference / 参考**: `references/journal-templates-guide.md` ### 12. `databases` — Scientific Database Search / 科学数据库检索 **Trigger / 触发**: `/academic databases` Unified search across 28+ scientific databases via MCP servers (arXiv, PubMed, OpenAlex, Semantic Scholar, FRED, ADS, etc.). Discipline-specific search strategies for CS, Medicine, Social Science, Physics, Economics. 通过 MCP 服务器统一检索 28+ 科学数据库。按学科定制搜索策略。 **Output / 产出**: Search results, BibTeX entries, citation graphs, deduplicated reference lists. **Reference / 参考**: `references/scientific-databases-guide.md` ### 13. `integrity` — Research Integrity Check / 研究诚信检查 **Trigger / 触发**: `/academic integrity` Four-layer verification: citation integrity (5-step), fact-checking protocol, methodology audit, cross-discipline compliance. Includes hallucination detection and plagiarism prevention. 四层验证：引文诚信、事实核查、方法论审计、跨学科合规。含幻觉检测和抄袭预防。 **Output / 产出**: Verification report with pass/fail per layer, actionable fixes. **Reference / 参考**: `references/research-integrity-guide.md` ### 14. `pre-submit` — Pre-Submission Review / 投前五维审查 **Trigger / 触发**: `/academic pre-submit` Five-dimension pre-submission quality review: Contribution, Technical Soundness, Writing, Experimental Rigor, Ethics. Scoring rubric (1-5 per dimension, ≥3.5 average to pass). 五维度投前质量审查：贡献、技术可靠性、写作、实验严谨性、伦理。每维度1-5分，均分≥3.5通过。 **Output / 产出**: Structured review report with scores, top-3 issues, pass/conditional/reject verdict. **Reference / 参考**: `references/pre-submission-review.md` ### 15. `benchmark` — Benchmark Paper Writing / 基准论文写作 **Trigger / 触发**: `/academic benchmark` Six-stage benchmark paper workflow: Scope → Data Collection → Metric Design → Baseline Selection → Analysis Framework → Writing. 六阶段基准论文工作流：范围→数据收集→指标设计→基线选择→分析框架→写作。 **Output / 产出**: Benchmark paper draft, dataset statistics, results tables, analysis framework. **Reference / 参考**: `references/benchmark-paper-template.md` ### 16. `reproduce` — Paper Reproduction / 论文复现 **Trigger / 触发**: `/academic reproduce` Six-phase reproduction: Paper Parsing → Implementation → Component Testing → Debugging → Sensitivity Analysis → Documentation. 六阶段复现：论文解析→实现→组件测试→调试→敏感性分析→文档化。 **Output / 产出**: Reproduction report, comparison table (paper vs. yours), sensitivity analysis. **Reference / 参考**: `references/paper-reproduction-guide.md` ### 17. `vibe` — AI Collaboration Rules / AI 协作研究 **Trigger / 触发**: `/academic vibe` Human-AI collaboration best practices: Vibe Coding (specify→verify→test), Vibe Writing (outline→expand→rewrite→polish), Vibe Figures (design→render→review). 人机协作最佳实践：Vibe Coding、Vibe Writing、Vibe Figures。含协作光谱和反模式。 **Output / 产出**: Collaboration workflow, anti-pattern warnings, quality checklist. **Reference / 参考**: `references/vibe-research-workflow.md` ### 18. `deep-read` — Deep Paper Reading / 论文精读 **Trigger / 触发**: `/academic deep-read` Four-pass systematic single-paper analysis: Bird's Eye (5 min) → Structural Decomposition (30-60 min) → Critical Analysis (60-120 min) → Internalization (30-60 min). Includes assumption audit, claim-evidence gap analysis, and one-page summary template. 四遍系统化单篇论文分析：全景扫描→结构拆解→批判分析→内化吸收。含假设审计、论点-证据差距分析、一页纸总结模板。 **Output / 产出**: Structured paper notes, critical analysis report, connection map to your work, one-page summary. **Reference / 参考**: `references/deep-reading-guide.md` ### 19. `data` — Research Data Processing / 研究数据处理 **Trigger / 触发**: `/academic data` Seven-stage data pipeline: Acquisition → Audit → Cleaning → Feature Engineering → EDA → Splitting & Versioning → Readiness Check. Cross-discipline patterns for CS, Medicine, Social Science, Economics, Physics. 七阶段数据流水线：获取→审计→清洗→特征工程→探索分析→划分与版本管理→就绪检查。覆盖CS、医学、社科、经济、物理。 **Output / 产出**: Cleaned dataset with provenance, audit report, EDA visualizations, split metadata. **Reference / 参考**: `references/data-processing-guide.md` ### 20. `tools` — Tool Registry & Recommendation / 工具注册与推荐 **Trigger / 触发**: `/academic tools` Curated scientific tool recommendations mapped to research stages and disciplines. Decision tree for tool selection: data acquisition → processing → EDA → modeling → visualization → writing. Covers 60+ tools across CS, biology, chemistry, medicine, physics, economics, social science. 精选科研工具推荐，按研究阶段和学科映射。工具选择决策树：数据获取→处理→探索→建模→可视化→写作。覆盖 60+ 工具。 **Output / 产出**: Tool recommendation for your specific stage + discipline + task. **Reference / 参考**: `references/tool-registry.md` ### 21. `hypothesis` — Hypothesis Generation & Critical Thinking / 假设生成与批判性思维 **Trigger / 触发**: `/academic hypothesis` Structured hypothesis generation from research gaps: identify assumptions, formulate testable hypotheses (H1, H2, H3), design falsification criteria, and evaluate hypothesis strength. Applies scientific critical thinking framework to avoid common reasoning fallacies. 从研究空白出发的结构化假设生成：识别假设、形成可检验假说、设计证伪标准、评估假设强度。应用科学批判性思维框架避免常见推理谬误。 **Output / 产出**: Hypothesis table (H# | Statement | Test | Falsification Criterion | Strength), reasoning chain validation report. **Reference / 参考**: `references/hypothesis-generation-guide.md` ### 22. `stat-analysis` — Statistical Analysis / 统计分析 **Trigger / 触发**: `/academic stat-analysis` Statistical analysis method selection and execution: hypothesis testing, effect sizes, confidence intervals, multiple comparison correction, power analysis. Integrates Python (scipy/statsmodels), R, and Stata tool recommendations. 统计分析方法选择与执行：假设检验、效应量、置信区间、多重比较校正、功效分析。集成 Python/R/Stata 工具推荐。 **Output / 产出**: Analysis plan (test selection rationale), execution code, results table with effect sizes and CIs. **Reference / 参考**: `references/statistical-analysis-guide.md` ### 23. `tool-deep` — Per-Package Deep Dive / 单包深度指南 **Trigger / 触发**: `/academic tool-deep <package-name>` In-depth usage guide for a specific research package: installation, core APIs, academic workflows, best practices, common pitfalls, and integration with the academic research pipeline. Covers 140+ packages across CS/ML, Biology, Chemistry, Physics, Engineering, Social Science, and more. 单个研究包的深度使用指南：安装、核心 API、学术工作流、最佳实践、常见陷阱、跨平台集成。覆盖 140+ 跨学科包。 **Output / 产出**: Package-specific guide with code examples, academic workflow templates, and discipline-specific tips.