academic-writing

学术写作与研究方法

学术写作、研究设计和学术交流的系统化方法。

研究设计基础

研究问题的制定

## Crafting a research question

### The FINER criteria
- **F**easible: Can you actually do this research?
- **I**nteresting: Does it matter to the field?
- **N**ovel: Does it add new knowledge?
- **E**thical: Can it be done ethically?
- **R**elevant: Does it address a real problem?

### Question types
| Type | Purpose | Example |
|------|---------|---------|
| Descriptive | Document phenomena | "What are the characteristics of X?" |
| Comparative | Compare groups/conditions | "How does X differ between groups?" |
| Correlational | Examine relationships | "Is there a relationship between X and Y?" |
| Causal | Establish causation | "Does X cause Y?" |
| Exploratory | Generate hypotheses | "What factors might explain X?" |

### Refining your question
Start broad → Narrow progressively

Draft 1: "How does social media affect politics?"
Draft 2: "How does Twitter use affect political polarization?"
Draft 3: "How does exposure to partisan Twitter accounts affect
         political attitude polarization among US adults?"
Draft 4: "Does increased exposure to ideologically homogeneous Twitter
         feeds increase affective polarization among politically
         engaged US adults aged 18-35?"

文献综述策略

## Systematic literature search

### Database selection by field
- **Multidisciplinary**: Web of Science, Scopus, Google Scholar
- **Social Sciences**: JSTOR, ProQuest, SSRN
- **Communication**: Communication Abstracts, ComAbstracts
- **Health**: PubMed, MEDLINE, CINAHL
- **Education**: ERIC
- **Business**: Business Source Complete

### Search strategy template
1. **Identify key concepts** from research question
2. **Generate synonyms** for each concept
3. **Combine with Boolean operators**

Example for: "social media political polarization"

Concept 1: social media
- OR: Twitter, Facebook, Instagram, "social networking sites",
      "online platforms", "digital media"

Concept 2: political
- OR: partisan, ideological, electoral, civic

Concept 3: polarization
- OR: division, extremism, "attitude change", radicalization

Combined search:
(Twitter OR Facebook OR "social media" OR "social networking")
AND (political OR partisan OR ideological)
AND (polarization OR division OR extremism)

### Inclusion/exclusion criteria
Document your criteria:
- Publication date range: [X to present]
- Languages: [English only / multiple]
- Publication types: [Peer-reviewed only / include preprints]
- Geographic scope: [Global / specific countries]
- Methodologies: [All / specific approaches]

### Managing your search
- Save searches to re-run later
- Export results to reference manager
- Track number of results at each stage
- Document date of each search

引文管理

# Zotero/reference manager integration patterns

# For automated citation workflows
CITATION_STYLES = {
    'apa7': 'American Psychological Association 7th edition',
    'mla9': 'Modern Language Association 9th edition',
    'chicago': 'Chicago Manual of Style 17th edition',
    'harvard': 'Harvard Reference format',
    'ieee': 'IEEE',
    'vancouver': 'Vancouver (medicine)',
}

# BibTeX entry template
BIBTEX_ARTICLE = """
@article{{{citekey},
    author = {{{author}}},
    title = {{{title}}},
    journal = {{{journal}}},
    year = {{{year}}},
    volume = {{{volume}}},
    number = {{{number}}},
    pages = {{{pages}}},
    doi = {{{doi}}}
}}
"""

# Common citation patterns by context
CITATION_CONTEXTS = {
    'introducing_concept': "According to Author (Year), ...",
    'supporting_claim': "Research has shown that X (Author, Year; Author, Year).",
    'contrasting': "While Author (Year) argues X, Author (Year) contends Y.",
    'methodology_reference': "Following the method developed by Author (Year), ...",
    'direct_quote': 'Author (Year) states that "exact quote" (p. X).',
}

论文结构与写作

IMRaD 结构（科学论文）

## Standard research paper sections

### Abstract (150-300 words typically)
- Background (1-2 sentences)
- Purpose/objective (1 sentence)
- Methods (2-3 sentences)
- Results (2-3 sentences)
- Conclusions (1-2 sentences)

### Introduction
**Funnel structure:**
1. Broad context - Why does this topic matter?
2. Narrow focus - What's the specific problem?
3. Gap identification - What don't we know?
4. Research question/hypothesis - What will you investigate?
5. Contribution preview - Why does this study matter?

### Literature Review
**Organize thematically, not chronologically:**
1. Theme 1: Key findings, debates, gaps
2. Theme 2: Key findings, debates, gaps
3. Theme 3: Key findings, debates, gaps
4. Synthesis: How themes connect to your study

### Methods
**The reproducibility test:** Could another researcher replicate your study from this section alone?

Include:
- Participants/sample (who, how selected, N)
- Materials/measures (what instruments, their validity)
- Procedure (step-by-step what happened)
- Analysis approach (statistical tests, qualitative methods)
- Ethical considerations (IRB, consent)

### Results
**Report, don't interpret:**
- Present findings systematically
- Use tables/figures for complex data
- Report effect sizes, not just p-values
- Address each research question/hypothesis

### Discussion
**Reverse funnel structure:**
1. Summary of key findings
2. Interpretation in context of literature
3. Theoretical implications
4. Practical implications
5. Limitations
6. Future research directions
7. Conclusion

学术写作风格

## Writing conventions

### Voice and tense
| Section | Tense | Example |
|---------|-------|---------|
| Abstract | Past (methods/results), Present (conclusions) | "We found... This suggests..." |
| Introduction | Present (established knowledge) | "Research shows..." |
| Methods | Past | "Participants completed..." |
| Results | Past | "Analysis revealed..." |
| Discussion | Present + Past | "These findings indicate... We found..." |

### Hedging language
Appropriate hedging (avoiding overclaiming):
- "This suggests that..." (not "This proves that...")
- "may be related to" (not "causes")
- "Results indicate..." (not "Results demonstrate conclusively...")
- "One possible explanation..." (not "The explanation...")

### Transition words by function
**Addition:** furthermore, moreover, additionally, in addition
**Contrast:** however, nevertheless, conversely, in contrast
**Cause/effect:** therefore, consequently, as a result, thus
**Example:** for instance, specifically, to illustrate
**Sequence:** first, subsequently, finally, meanwhile
**Summary:** in summary, overall, in conclusion

### Paragraphs
Each paragraph should:
1. Begin with a topic sentence
2. Contain one main idea
3. Include supporting evidence
4. Connect to adjacent paragraphs
5. Average 100-200 words (varies by field)

常见写作问题

## Issues to avoid

### Wordiness
❌ "It is important to note that the results demonstrate..."
✅ "Results demonstrate..."

❌ "In order to investigate..."
✅ "To investigate..."

❌ "A total of 50 participants..."
✅ "Fifty participants..."

### Weak verbs
❌ "The study was conducted by the researchers"
✅ "The researchers conducted the study"

❌ "There was a significant difference found"
✅ "We found a significant difference"

### Vague language
❌ "Several studies have shown..."
✅ "Three studies (Author, Year; Author, Year; Author, Year) showed..."

❌ "The results were significant"
✅ "The results were statistically significant (p < .05, d = 0.45)"

### Unnecessary jargon
- Define technical terms on first use
- Use simpler words when equally precise
- Consider your audience's expertise level

### Citation problems
❌ Citing secondary sources without noting
✅ "(Author, Year, as cited in Author, Year)"

❌ String citations without synthesis
✅ Group citations that make the same point; distinguish those that differ

同行评审与修订

回复审稿人

## Response letter template

Dear Editor and Reviewers,

Thank you for the thoughtful feedback on our manuscript titled "[Title]"
(Manuscript ID: [Number]). We have carefully considered all comments and
revised the manuscript accordingly. Below, we provide point-by-point
responses to each comment.

---

## Reviewer 1

### Comment 1.1
[Quote or paraphrase the reviewer's comment]

**Response:**
[Your response explaining what you did]

**Changes made:**
[Quote the specific changes with page/line numbers]
"New text here..." (p. X, lines XX-XX)

### Comment 1.2
[Continue for each comment]

---

## Reviewer 2
[Same format]

---

We believe these revisions have substantially strengthened the manuscript
and hope you will find it suitable for publication in [Journal Name].

Sincerely,
[Authors]

处理批评意见

## Types of reviewer feedback

### Must address
- Methodological concerns
- Missing literature
- Unclear writing
- Logical gaps in argument
- Statistical errors

### Negotiate carefully
- Requests for additional analyses
- Suggestions to restructure
- Disagreements on interpretation

### Pushback appropriately
- Requests outside scope
- Misunderstandings of your argument
- Contradictory advice from reviewers

### Response strategies
**Agreement:** "We thank the reviewer for this suggestion. We have [action]."
**Partial agreement:** "We appreciate this point. While [acknowledge validity], we [explain your approach]. However, we have [partial accommodation]."
**Respectful disagreement:** "We thank the reviewer for raising this issue. We have considered this carefully; however, [explanation]. We hope the reviewer will find this reasoning persuasive."

基金申请书

申请书结构（NSF/NIH 风格）

## Standard grant components

### Specific aims (1 page)
**Opening paragraph:** What's the problem? Why does it matter?
**Gap statement:** What's missing from current understanding?
**Long-term goal:** Your research program vision
**Objective:** What this specific project will accomplish
**Central hypothesis:** Your testable prediction
**Aims:** 2-4 specific, achievable objectives
**Impact statement:** Why funding this matters

### Significance (2-3 pages)
- Importance of the problem
- Gaps in current knowledge
- How your work advances the field
- Potential impact if successful

### Innovation (1 page)
- What's new about your approach?
- Conceptual innovation
- Methodological innovation
- Technical innovation

### Approach (6-12 pages)
For each aim:
- Rationale
- Preliminary data (if any)
- Research design
- Methods
- Expected outcomes
- Potential problems and alternatives
- Timeline

### Broader impacts
- Training opportunities
- Dissemination plans
- Benefits to society
- Diversity and inclusion

预算说明

## Budget categories

### Personnel
- PI salary and effort (% time)
- Co-investigators
- Postdocs (salary + benefits)
- Graduate students (stipend + tuition + benefits)
- Undergraduate researchers
- Technical staff

### Equipment
- Items over $5,000 (check sponsor threshold)
- Justify necessity for project

### Supplies
- Lab consumables
- Software licenses
- Participant payments

### Travel
- Conference presentations
- Data collection sites
- Collaborator visits

### Other direct costs
- Publication costs
- Participant incentives
- Transcription services
- Equipment maintenance

### Indirect costs (F&A, facilities and administrative)
- Each institution negotiates an F&A rate with its cognizant federal agency.
- The rate applies to **modified total direct costs (MTDC)**, not raw total
  direct costs, exclude equipment >$5K, tuition, participant support,
  subaward portions over $25K, and a few other categories per the
  institution's negotiated agreement.
- R1 universities commonly negotiate 55-70% of MTDC; smaller institutions
  often run lower. Always confirm the current rate with your sponsored
  programs office before budgeting.

投稿策略

期刊选择

## Evaluating journals

### Quality indicators
- Impact factor (use cautiously)
- Acceptance rate
- Review time
- Time to publication
- Reputation in your field
- Indexing (Web of Science, Scopus)

### Fit indicators
- Scope alignment
- Audience match
- Article type (empirical, theoretical, review)
- Open access route (see below)

### Open access routes
- **Gold OA**, published OA in a fully OA journal; usually requires an
  Article Processing Charge (APC), often $1,000-$3,000+.
- **Green OA**, closed-access journal but author self-archives an
  accepted manuscript in a repository (institutional, arXiv, etc.) under
  the publisher's allowed embargo.
- **Diamond / Platinum OA**, fully OA journal with no APC; supported
  by a sponsor (society, university, consortium). Increasingly common
  in social sciences and humanities.
- **Hybrid**, a closed-access journal that lets individual articles
  go OA for an APC. Note: hybrid is *excluded* from Plan S compliance
  and from many funder OA mandates, verify before paying an APC.
- **Bronze**, free to read but no formal OA license; the publisher
  can revoke access. Treat as not-actually-OA for compliance purposes.

### Funder mandates (US + Europe, as of 2026)
- **OSTP Nelson Memo (US)**, federally-funded research must be made
  immediately publicly accessible upon publication, with no embargo.
  Implementation deadline was **end of 2025**, so as of 2026 this is
  in effect for all major US federal funders (NIH, NSF, DOE, NASA, etc.).
- **Plan S (cOAlition S)**, European funders (Wellcome, UKRI, ANR,
  many others) require immediate Gold or Diamond OA, or Green OA
  with no embargo and a CC-BY license. Hybrid is not compliant.
- Verify the funder's most recent policy before submission, accepted
  routes shift, and some funders cap APCs.

### Red flags (predatory journals)
- Aggressive email solicitation
- Rapid "peer review" (days)
- Unclear editorial board
- Not indexed in major databases
- "Pay to publish" with no clear OA model
- Poor website quality

### Resources for vetting journals + ethics
- **Think. Check. Submit.** (thinkchecksubmit.org), checklist for legitimacy.
- **DOAJ** (Directory of Open Access Journals), vetted OA-journal index.
- **Journal Citation Reports**, paid Web of Science product for impact factor.
- **Cabells Predatory Reports**, subscription-based replacement for the
  original Beall's List, which was taken down in 2017 (only stale mirrors
  remain, don't rely on them as a current source of truth).
- **Retraction Watch Database**, searchable record of retractions and
  publisher misconduct; useful for vetting authors and journals.
- **COPE** (Committee on Publication Ethics, publicationethics.org),
  authoritative guidelines for editors and authors on ethics and integrity.
- **ICMJE** (International Committee of Medical Journal Editors,
  icmje.org/recommendations), defines authorship criteria, conflict-of-
  interest disclosure, and reporting standards used across many fields.

投稿附信模板

Dear Dr. [Editor's name],

We are pleased to submit our manuscript titled "[Full title]" for
consideration as a [article type] in [Journal Name].

**Summary (2-3 sentences):**
[What you did and what you found]

**Significance (2-3 sentences):**
[Why this matters for the journal's readership]

**Fit statement:**
[Why this journal specifically]

**Declarations:**
- This manuscript is original and not under consideration elsewhere
- All authors have approved the submission
- [Conflict of interest statement]
- [Funding acknowledgment]

**Suggested reviewers (if requested):**
1. [Name, affiliation, email] - Expert in [relevant area]
2. [Name, affiliation, email] - Expert in [relevant area]
3. [Name, affiliation, email] - Expert in [relevant area]

**Reviewers to exclude (if any):**
[Name] - [Brief, professional reason]

We look forward to your response.

Sincerely,
[Corresponding author name]
[Title, affiliation]
[Contact information]

预印本与持久标识符

预印本服务器（发布工作版本，以便更早获得引用）

预印本是作者提交、在正式同行评审前发布的稿件。如今，大多数期刊都允许（或积极鼓励）发布预印本；如果不确定，请在发布前查阅目标期刊的政策。

服务器	领域	备注
arXiv (arxiv.org)	物理学、数学、计算机科学、定量生物学、经济学、统计学	历史最悠久、规模最大；按分区进行审核。
SSRN (ssrn.com)	社会科学、法律、经济学、金融学	自 2016 年起由 Elsevier 所有，部分作者更倾向于选择 SocArXiv 作为替代方案。
bioRxiv (biorxiv.org)	生物学	由 Cold Spring Harbor Laboratory 运营；引用广泛。
medRxiv (medrxiv.org)	卫生科学、医学	对安全性声明进行宽松审核；未经预注册，不要发布临床试验结果的预印本。
EarthArXiv	地球科学、行星科学、环境科学	依托 OSF 基础设施，由社区运营。
PsyArXiv	心理学与行为科学	由 OSF 托管。
SocArXiv	社会学与社会科学	由 OSF 托管；是独立于 Elsevier 的 SSRN 替代方案。
EngrXiv	工程学	由 OSF 托管。

**为什么发布预印本：**更早为优先权主张留下时间戳，可在期刊接收数月前获得引用，并在最终评审前获得更广泛的反馈。（2026 年 OSTP Nelson Memo 规定，联邦资助的美国研究成果经同行评审的版本在发表时须立即向公众开放；这并不是预印本发布要求。预印本发布是另一项自愿选择，可以补充 Nelson Memo 合规要求，但不能代替合规。）

持久标识符：获取 ORCID 并使用 DOI

• ORCID iD (orcid.org)：免费的 16 位标识符，可在不同出版商、资助方和机构之间明确关联你的身份。大多数主要期刊和资助方都要求提供。请将其关联到你的简历、稿件、数据集、软件和基金申请。
• DOI (doi.org)：学术成果的持久标识符。期刊通常会在稿件接收后自动分配 DOI；你也可以通过 Zenodo (zenodo.org) 或所在机构的知识库，为自己的数据集、代码、海报和预印本创建 DOI，便于引用非期刊成果。
• ROR (ror.org)：研究机构的持久标识符，资助方和出版商用它来区分名称相似的机构。

研究伦理

伦理考量检查清单

## Before starting research

### Human subjects
- [ ] IRB/ethics board approval obtained
- [ ] Informed consent procedures established
- [ ] Vulnerable populations identified and protected
- [ ] Privacy and confidentiality measures in place
- [ ] Data security plan established
- [ ] Risk/benefit ratio acceptable

### Data management
- [ ] Data management plan created
- [ ] Secure storage arranged
- [ ] Sharing/archiving plans documented
- [ ] Retention period determined
- [ ] Destruction procedures established

### Authorship
- [ ] Contribution criteria discussed (use ICMJE definition: substantial
      contribution + drafting/revising + final approval + accountability)
- [ ] Author order agreed upon
- [ ] All contributors will meet authorship criteria
- [ ] Acknowledgments planned for non-author contributors
- [ ] **CRediT taxonomy** (credit.niso.org) roles assigned for each
      author, most major journals now require or recommend CRediT,
      which has 14 standardized contributor roles (Conceptualization,
      Methodology, Software, Validation, Formal analysis, Investigation,
      Resources, Data curation, Writing - original draft, Writing -
      review & editing, Visualization, Supervision, Project administration,
      Funding acquisition).

### Conflicts of interest
- [ ] Financial conflicts identified
- [ ] Non-financial conflicts identified
- [ ] Disclosure plan established

### Reproducibility
- [ ] Analysis plan pre-registered on **OSF** (osf.io) or **AsPredicted**
      (aspredicted.org) before data collection, distinguishes
      confirmatory from exploratory analyses and protects against
      HARKing. Check whether the target journal accepts **Registered
      Reports** (peer review of the design *before* data collection,
      with in-principle acceptance regardless of results), this format
      now exists at 350+ journals across psychology, biomedical, and
      social sciences as of 2026.
- [ ] Code will be shared (consider a GitHub release + Zenodo DOI for
      versioned, citable code archives)
- [ ] Data will be shared (FAIR principles; institutional repository,
      Dryad, Figshare, or domain-specific repository as appropriate)
- [ ] Materials will be shared

避免研究不端行为

## Types of misconduct

### Fabrication
- Making up data or results
- Never acceptable under any circumstances

### Falsification
- Manipulating data, equipment, or processes
- Selectively omitting data to change conclusions
- Image manipulation beyond acceptable adjustment

### Plagiarism
- Using others' words without attribution
- Self-plagiarism (reusing own published work without acknowledgment)
- Paraphrasing too closely

### Other questionable practices
- P-hacking (running multiple analyses until significant)
- HARKing (hypothesizing after results known)
- Salami slicing (fragmenting one study into many papers)
- Gift/ghost authorship
- Selective reporting of results

在学术写作中使用 AI / LLM

LLM 辅助写作是 2024—2026 年学术出版领域最重要的伦理议题。各大期刊、资助方和伦理机构都已在这一时期发布政策，且政策仍在日趋严格。请将下文内容视为最低要求，而非最高标准：投稿前，请阅读目标期刊当前的投稿指南以及资助方最新的政策。

截至 2026 年普遍禁止的事项

• 由大模型署名。 ICMJE、COPE、Nature、Science、NEJM、Cell、JAMA、Lancet 以及各大大学出版社都明确禁止将大模型（ChatGPT、Claude、Gemini 等）列为共同作者。大模型无法满足作者身份所要求的责任承担和批准标准。应在致谢部分或方法／披露声明中说明，绝不能将其列入作者署名。
• 生成伪造的引文。 有充分记录表明，大模型会生成看似可信、实际却是伪造的 DOI、页码，甚至不存在的作者／标题组合。稿件中的每条引文都必须对照来源核实；自 2023 年以来，由大模型生成的虚假引文一直是导致撤稿的主要原因之一。
• 生成数据、结果或图像。 合成实验数据、伪造图表，或使用生成式 AI“补全”实际并未获得的结果，都属于 COPE 定义的研究不端行为。
• 未披露实质性使用。 大多数期刊要求披露任何超出简单拼写／语法检查辅助的大模型使用情况。未披露有实质意义的使用可能成为撤稿的理由。

通常允许的事项（需披露）

• 构思创意和拟定大纲
• 润色语言和纠正语法
• 翻译自己的文字
• 为分析生成代码（需明确测试）
• 总结自己的笔记或转录文本
• 生成模板化章节（求职信、IRB 用语），之后由你进行事实核查并对内容负责

披露用语（模板）

大多数期刊希望在方法或致谢声明中注明所用工具、版本（如有）及用途。示例：

在准备本研究期间，作者使用了 [工具名称和确切版本，例如 GPT-5.6 Sol、Claude Opus 5、Gemini 3.7 Flash]，用于 [具体用途，例如润色引言语言；为第 3.2 节的聚类分析编写代码]。使用该工具后，作者审阅并按需编辑了相关内容，并对出版物内容承担全部责任。

请根据目标期刊要求的确切措辞进行调整；Elsevier、Springer Nature、Wiley、Taylor & Francis、IEEE 和 ACM 都各自发布了偏好的表述。

投稿前披露清单

• 阅读目标期刊当前投稿指南中的 AI／大模型政策（政策每年会更新多次）。
• 阅读资助方的政策；部分资助方（Wellcome、NSF）的规定比期刊更严格。
• 使用期刊偏好的表述添加披露声明。
• 手动核实每条引文；对于任何 DOI、作者名单、页码范围或引用段落，都不要轻信大模型的输出。
• 核实每一项数值主张、日期和具名事实。
• 通读整篇稿件，确保主导稿件的是你的表达方式和论证结构，而不是大模型。
• 如果使用大模型编写代码或进行分析，请使用随机种子重新运行，并核实报告的所有数字都能复现。

检测工具（可靠性有限）

GPTZero、Turnitin AI Detection、Originality.ai 等类似工具已被证实，对非英语母语写作者以及由人类正常撰写的技术文本有较高的误报率；对经过少量编辑的当前模型输出则有较高的漏报率。Pangram Labs（pangram.com）发布了更严格的基准测试，在大致相当的设置下通常优于较早的检测器，但它仍不能作为最终裁定依据；当作者对大模型输出进行大量编辑时，它也存在同样的根本局限。检测结果绝不能单独作为认定不端行为的依据；越来越多曾将其作为把关信号的期刊正在撤回这种做法。应将检测结果视为需要与作者进一步讨论的提示，而非单独的证据。

效率与工作流

写作习惯

## Sustainable academic writing

### Daily writing practice
- Write at same time daily (habit formation)
- Start with minimum viable goal (e.g., 30 minutes)
- Track progress (word counts, time)
- Protect writing time from meetings/email

### Managing large projects
1. Break into smallest possible tasks
2. Set deadlines for each component
3. Schedule regular writing blocks
4. Build in buffer time
5. Get feedback early and often

### Overcoming blocks
- Start with easiest section
- Write a "shitty first draft" (Anne Lamott)
- Use placeholders for citations/data
- Talk through ideas with colleague
- Change environment
- Return to outline/structure

学术工具

用途	工具
参考文献管理	Zotero（推荐；开源，可跨设备同步）、Paperpile（以网页为先，集成 Google Docs）、EndNote、Citavi。**关于 Mendeley 的说明：**Elsevier 于 2022 年 9 月宣布停止提供 Mendeley Desktop 下载，之后又恢复了下载。Mendeley Desktop 仍是旧版软件，不再开发新功能；Mendeley Reference Manager（网页/云端）才是目前积极维护的产品。新用户使用 Zotero 或 Paperpile 会更合适。
写作	Scrivener、Overleaf（LaTeX 协作）、Word、Typst（较新的 LaTeX 替代方案）
协作	Google Docs、Overleaf、Hedgedoc
版本控制	Git / GitHub / GitLab（真正的版本控制）。Word 的修订跟踪属于协作编辑，而非版本控制，它不具备分支、合并、历史重写和原子提交功能。
任务管理	Todoist、Notion、Trello
专注	Forest、Freedom、Cold Turkey
分析	R、Python、SPSS、Stata、NVivo、Julia
可视化	R/ggplot2、Python/matplotlib + plotly、Observable、Tableau
大模型辅助写作	使用其中任何工具之前，请参阅上文“AI / 大模型在学术写作中的使用”部分，了解披露要求。