Chinese Thesis (latex-thesis-zh)

Chinese Thesis (latex-thesis-zh)

LaTeX assistant for Chinese doctoral/master theses.

Overview

The latex-thesis-zh skill provides modular support for Chinese thesis writing. Modules are independently callable, but structure mapping should run first for full reviews or multi-file theses.

Using the Skill in Claude Code

This skill is designed to work with Claude Code and similar AI assistants. Simply mention the relevant trigger words in your conversation, and the assistant will activate the appropriate module.

Argument Conventions

Provide clear inputs in your request:

• Main .tex path (required for tool execution)
• Target scope (chapter/section or full thesis)
• Module choice (structure / compile / GB/T / template / etc.)

If any of these are missing or ambiguous, the assistant will ask for clarification instead of guessing.

Execution Guardrails

• Tools/scripts run only when you explicitly request execution.
• Destructive operations (--clean, --clean-all) require explicit confirmation.

Example Usage

Compile thesis:

Compile thesis.tex using xelatex with biber

Map structure:

Map the structure of my thesis to check completeness

Check GB/T format:

Check thesis.tex for GB/T 7714-2015 compliance

Reduce AI traces:

Reduce AI writing traces in the introduction chapter

Trigger Words

Output Protocol

All suggestions must use a diff-comment style and include fixed fields:

• Severity: Critical / Major / Minor
• Priority: P0 / P1 / P2

Minimal template:

% <模块>（第<N>行）[Severity: <Critical|Major|Minor>] [Priority: <P0|P1|P2>]: <问题概述>
% 原文：...
% 修改后：...
% 理由：...
% ⚠️ 【待补证】：<需要证据/数据时标记>

If a tool fails (missing script/tool or invalid path), respond with an error comment and a safe next step.

Modules

Compile Module

XeLaTeX-focused compilation for Chinese documents.

Tools (aligned with VS Code LaTeX Workshop):

Recipes:

Usage in Claude Code:

Ask the assistant to compile your thesis:

Compile thesis.tex using xelatex

Compile thesis.tex with xelatex and biber workflow

Clean auxiliary files for thesis.tex

The assistant will execute the appropriate compilation commands.

Failure handling:

• Missing LaTeX tools: install TeX Live/MiKTeX and ensure PATH is set
• Missing file/script: verify working directory and scripts/ path
• Compilation error: summarize the first error and request the relevant log snippet

Structure Mapping Module

Analyze multi-file thesis structure. Run this first for full reviews or multi-file theses.

Usage in Claude Code:

Map the structure of thesis.tex

Analyze the structure of my thesis to check for missing sections

Thesis Structure Requirements:

GB/T Format Check Module

Verify GB/T 7714-2015 compliance.

Usage in Claude Code:

Check thesis.tex for GB/T format compliance

Check thesis.tex format with strict mode

Checks:

• Bibliography format (biblatex-gb7714-2015)
• Figure/table caption format
• Equation numbering
• Section heading styles

Academic Expression Module

Detect colloquial expressions and suggest academic alternatives.

Colloquial → Academic Examples:

Usage in Claude Code:

Improve academic expression in Chapter 2

Replace colloquial phrases in my abstract

Logic & Methodology Module

Ensure logical flow between paragraphs and strengthen methodological rigor.

AXES Model for Paragraph Coherence:

Transition Signals (Chinese):

Common Issues:

Usage in Claude Code:

Check logical coherence in my introduction chapter

Analyze methodological depth in the methods chapter

Long Sentence Analysis Module

Trigger: Sentences >60 characters OR >3 clauses

Output: Core extraction, modifier analysis, rewrite suggestions

Usage in Claude Code:

Split long sentences in the introduction chapter

Analyze sentences longer than 60 characters in Chapter 3

Bibliography Module

Usage in Claude Code:

Verify refs.bib for GB7714 standard compliance

Check refs.bib against thesis.tex for format errors

Behind the scenes, the assistant uses:

python scripts/verify_bib.py references.bib
python scripts/verify_bib.py references.bib --standard gb7714

Template Detection Module

Usage in Claude Code:

Detect the template used in thesis.tex

Identify which university template my thesis is using

The detector also returns a key requirements summary extracted from references/UNIVERSITIES/.

Supported Templates:

University Guidelines (References):

• Yanshan University (2024 thesis writing guidelines): see references/UNIVERSITIES/yanshan.md (reference only; template detection still relies on \documentclass{...}).

De-AI Polishing Module

Reduce AI writing traces while preserving LaTeX syntax and technical accuracy.

Input Requirements:

1. Source type (required): LaTeX / Typst
2. Section (required): Abstract / Introduction / Related Work / Methods / Experiments / Results / Discussion / Conclusion / Other
3. Source snippet (required): paste directly, keep indentation and line breaks

Usage in Claude Code:

Interactive editing (single section):

Analyze AI writing traces in the introduction section of thesis.tex

Batch processing (chapter or full document):

Process chapter3/introduction.tex to reduce AI traces

Process all sections in thesis.tex to reduce AI traces

AI trace density check:

Analyze AI trace density across all sections in thesis.tex

Workflow:

1. Syntax structure identification (preserve all LaTeX/Typst constructs):
   • Commands: \command{...}, \command[...]{}
   • References: \cite{}, \ref{}, \label{}, \eqref{}, \autoref{}
   • Environments: \begin{...}...\end{...}
   • Math: $...$, \[...\], equation/align environments
   • Custom macros (unchanged by default)
2. AI pattern detection:
   • Empty phrases: "significant", "comprehensive", "effective", "important"
   • Over-confident: "obviously", "necessarily", "completely", "clearly"
   • Mechanical structures: empty three-part parallelisms
   • Template expressions: "in recent years", "more and more"
3. Text rewriting (visible text only):
   • Split long sentences (>50 words)
   • Adjust word order for natural flow
   • Replace vague claims with specific statements
   • Delete redundant phrases
   • Add required subjects without introducing new facts
4. Output generation:
   • A. Rewritten source code (minimal invasive edits)
   • B. Change summary (3-10 bullets)
   • C. Pending verification marks (claims requiring evidence)

Hard Constraints:

• Never modify: \cite{}, \ref{}, \label{}, math environments
• Never add: new data, metrics, comparisons, contributions, experimental settings, citation numbers, or bib keys
• Only modify: visible paragraph text, section titles, caption text

Output Format:

% ============================================================
% DE-AI EDITING (Line 23 - Introduction)
% ============================================================
% Original: This method achieves significant performance improvement.
% Revised: The proposed method improves performance in the experiments.
%
% Changes:
% 1. Removed vague phrase: "significant" → deleted
% 2. Kept the claim without adding new metrics or baselines
%
% ⚠️ [PENDING VERIFICATION]: Add exact metrics/baselines only if supported by data
% ============================================================

\section{Introduction}
The proposed method improves performance in the experiments...

Section-Specific Guidelines:

Reference: references/DEAI_GUIDE.md

Title Optimization Module

Generate and optimize thesis titles following GB/T 7713.1-2006 standards and international best practices.

Key Principles

Based on GB/T 7713.1-2006 and international academic standards:

1. 简洁性 (Conciseness): Remove "关于...的研究", "...的探索", "新型", "改进的"
2. 可搜索性 (Searchability): Place key terms (方法+问题) in first 20 characters
3. 长度 (Length): Optimal 15-25 characters; acceptable 10-30 characters
4. 具体性 (Specificity): Use concrete method/problem names, avoid vague terms
5. 规范性 (Compliance): Follow thesis title standards, avoid obscure abbreviations

Quality Scoring

Each title receives a score (0-100) based on:

• 简洁性 (25%): No ineffective words
• 可搜索性 (30%): Key terms in first 20 characters
• 长度 (15%): Within optimal range
• 具体性 (20%): Concrete vs vague terms
• 规范性 (10%): No obscure abbreviations

Usage in Claude Code

Check existing title:

检查我的论文标题质量

Generate title candidates:

根据摘要生成论文标题候选方案

Optimize existing title:

优化我的学位论文标题，符合国标规范

The assistant will analyze your title and provide:

• Quality score with breakdown (质量评分及细分)
• Specific issues detected (检测到的问题)
• Multiple improved candidates with Chinese-English pairs (中英文对照的改进方案)
• Suggested LaTeX code (建议的 LaTeX 代码)

Title Generation Workflow

1. 内容分析 (Content Analysis): Extract problem, method, domain from abstract/introduction
2. 关键词提取 (Keyword Extraction): Identify 3-5 core keywords
3. 模板选择 (Template Selection): Choose appropriate pattern
4. 候选生成 (Candidate Generation): Create 3-5 variants
5. 质量评分 (Quality Scoring): Rank candidates by score

Common Title Patterns

Ineffective Words to Remove

Good vs Bad Examples

好：工业控制系统时间序列预测的Transformer方法
差：关于基于Transformer的工业控制系统时间序列预测的研究

好：图神经网络故障检测方法及其工业应用
差：新型改进的基于图神经网络的故障检测方法研究

好：注意力机制的多变量时间序列预测方法
差：基于注意力机制的改进型多变量时间序列预测模型研究

University-Specific Requirements

清华大学 (Tsinghua University):

• 中文标题：不超过 36 个汉字
• 避免使用缩写和公式

北京大学 (Peking University):

• 中文标题：简明扼要，一般不超过 25 字
• 可使用副标题（用破折号分隔）

通用要求 (General Requirements):

• 遵循 GB/T 7713.1-2006 规范
• 中文标题：15-25 字为宜
• 英文标题：对应翻译，注意冠词和介词

Chinese-English Title Pairs

Best Practices

1. 关键词前置: 方法+问题放在前 20 字
2. 具体明确: "Transformer" > "深度学习" > "机器学习"
3. 删除冗余: 去掉"关于"、"研究"、"新型"、"基于"
4. 控制长度: 目标 15-25 字（中文）
5. 测试可搜索性: 用这些关键词能找到你的论文吗？
6. 避免生僻: 除非是广泛认可的术语（AI、LSTM、CNN）
7. 符合规范: 遵循学校模板和 GB/T 7713.1-2006 标准

References

• GB/T 7713.1-2006: 学位论文编写规则
• references/GB_STANDARD.md: 国标格式规范
• references/UNIVERSITIES/: 学校模板指南

Workflow Suggestions

Daily Writing

Ask the assistant:

Compile thesis.tex using xelatex

Chapter Completion

Ask the assistant:

Compile thesis.tex with xelatex and biber, then verify refs.bib for GB7714 compliance

Final Submission

Ask the assistant to perform a complete review:

Please perform a complete thesis review:
1. Map the structure of thesis.tex
2. Compile with xelatex and biber
3. Check refs.bib for GB7714 compliance
4. Check consistency across chapters
5. Clean auxiliary files

Recommended Workflows

Full Thesis Review (Multi-file)

1. Structure mapping (first)
2. GB/T format check
3. Template detection + verify key requirements
4. De-AI polishing (intro + related work)
5. Academic expression cleanup
6. Long sentence analysis
7. Bibliography verification

Pre-Defense Quick Pass

1. Compile with xelatex + biber
2. Bibliography verification
3. GB/T format check (strict)

Common Issues

Chinese Font Missing

\setCJKmainfont{SimSun}  % Windows
\setCJKmainfont{STSong}  % macOS

Bibliography Format Incorrect

\usepackage[backend=biber,style=gb7714-2015]{biblatex}