Skip to content

Dedicated Guide to Writing the Introduction (Chapter 1)

For writing and diagnosing the introduction of a Chinese doctoral/master's degree thesis. Diagnostic entry points:

bash
uv run python $SKILL_DIR/scripts/analyze_literature.py main.tex --intro-citations --bib refs.bib
uv run python $SKILL_DIR/scripts/analyze_logic.py main.tex --intro-mainline

All thresholds here are common criteria from multiple university rules and writing practice, with sources noted in each section. The latest rules of the author's graduate school take precedence; command-line options can override defaults.

Six-Section Funnel Skeleton and Length Allocation

The introduction should total about 10,000 Chinese characters and no more than 20 pages (expert consensus: “get to the point; 20 pages are enough”).

SectionFunctionReference Share
Research background and significanceDomain context -> engineering/theoretical value, ending at “why this is worth doing”~15%
Domestic and international research statusThematic review + evolution path + visual synthesis~45%
Existing problems/key challengesDerive unresolved problems from review boundaries~10%
Research contents and innovationsScientific problem -> research content -> innovation four-way closure~20%
Research method/technical route (optional by university)Overall solution and technical-route figure~5%
Thesis organizationProgressive relationship among chapters, not a mechanical title list~5%

The opening paragraph must complete a three-level funnel: domain background -> technical bottleneck -> what this thesis addresses. Do not begin directly with terminology or system details (script check L-FUN).

Citation Quota (B1/B4 Defaults and Sources)

MetricDefaultSource
Unique introduction citations120-160 or more (doctoral level)Several university rules, including Beijing University of Technology and Harbin Institute of Technology, require >=100 across the thesis; the introduction carries most citations
Latest-three-year share>=30%Common strict criterion; the broad criterion is latest-five-year share >=1/3 and must include latest-two-year literature
Latest-five-year share>=50%Same as above
Foreign-language share>=1/2 (whole-thesis scope)Rules at Beijing University of Technology and others; script does not yet check, review manually

Literature-selection proportions, used when collaborating with bib-search-citation:

  • Each thematic cluster = 1-2 foundational works + 2-3 works from the latest three years, balancing domestic and international sources;
  • Discuss multiple works by the same author/team separately, comparing method evolution and limitations instead of placing the whole cluster in one \cite (script checks B2/B3);
  • Prefer primary literature: cite the original work being discussed, not a secondary retelling;
  • Textbooks, newspapers, and unpublished reports should generally not support arguments;
  • When citations are insufficient, search only the user's .bib or a real literature database; never fabricate entries.

Organization and Visualization of Domestic/International Research Status (B5)

Choose one organization and declare it explicitly:

  1. Separate domestic/international discussion: foreign then domestic, or the reverse, followed by comparison of differences and connections (general-specific-general);
  2. Thematic mixing: open with “this section is organized by theme; domestic and international work is compared within each theme” (script check L-DOM: a title containing “domestic and international” must fulfill one of these forms).

Within each thematic cluster, follow 共识 -> 分歧 -> 局限 -> 空白; end with a comparison/concession/ limitation sentence (literature module A2). Order from distant to close relevance.

Visual synthesis (at least one; script check B5):

  • Research-evolution timeline: time on the horizontal axis, divided into phases such as emergence/development/deepening, with 2-3 representative works and landmark turns per phase;
  • Literature-comparison matrix: rows = representative works; columns = method route / key assumption / scope / limitation; close with a general-specific-general paragraph after the table;
  • Note: a Gantt chart belongs to proposal scheduling and does not go in the introduction. Use a research-evolution or technical-route figure when a route must be shown.

Three prohibitions (common review anti-patterns): lecture-style lists of theoretical schools; subjective straw-man criticism of prior work; vague sources or mixing prior work with the author's own work.

Refining Scientific Problems (L-SCI)

A scientific problem must contain all three elements: object, problem, method. Sentence template:

text
针对〈对象〉的〈现象/瓶颈〉,研究〈科学问题〉,拟采用〈方法路径〉。

Decision table:

FormExampleJudgment
Short noun phrase“多速率状态重叠”Invalid: only a topic term, with neither object nor method; script match
Technical task“开发一套软测量系统”Invalid: an engineering task, not a scientific problem
Valid problem sentence“针对多速率异步采样的烧成过程,研究低频质量条件如何调制高频动态特征,拟采用条件特征调制机制”Valid: object + problem + method, testable

Requirement: one clear, testable sentence that can guide data collection and verification. Keep one topic and decompose it top-down with a hypothesis tree; do not overreach.

Four-Way Closure (L-MAP)

Scientific problem <-> research content <-> innovation <-> chapter arrangement should map one to one. Recommended practice:

  • include a “scientific problem-research content-innovation-corresponding chapter” mapping table in the introduction;
  • when counts differ, state the reason in the body, such as “system implementation is an engineering-validation contribution and is not stated as equal to method innovations”; the script then degrades to Info;
  • answer every innovation in the conclusion chapter (logic module C3).

Holistic Writing Order

  1. Theme first: establish the thesis's single mainline sentence before selecting material; delete content unrelated to the research problem;
  2. Write body first, introduction later: refine background, status, and significance after gaining a holistic understanding of the research;
  3. Organize the review around this thesis's innovations, clearly explaining the “shoulders of predecessors” to position the contribution;
  4. Describe progressive chapter relationships (how Chapter N output becomes Chapter N+1 input), not chapter titles one by one.
latex
% 绪论引用(chapters/chapter1.tex:55)[Severity: Minor] [Priority: P2]: [Script] B3 前缀“zhao”文献 5 篇,且在第55行整簇共引
% 建议:拆开分述并比较该团队方法的演进与局限;若为不同作者请人工复核后忽略。

Further Reading

Released under the MIT License.