Back to Details

paper-analyzer

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Academic Paper Analyzer - 学术论文深度解析

Academic Paper Analyzer - In-depth Analysis of Academic Papers

核心能力

Core Capabilities

  • MinerU Cloud API 高精度 PDF 解析
  • 自动提取图片、表格、LaTeX 公式
  • 多种写作风格：故事型 / 学术型 / 精炼型
  • 可选公式讲解：插入公式图片并详细解读
  • 可选代码分析：结合 GitHub 开源代码讲解
  • 输出 Markdown + HTML（base64 嵌入图片）
  • MinerU Cloud API High-precision PDF Parsing
  • Automatically extract images, tables, and LaTeX formulas
  • Multiple Writing Styles: Storytelling / Academic / Concise
  • Optional Formula Explanations: Insert formula images and provide detailed interpretations
  • Optional Code Analysis: Explain in combination with GitHub open-source code
  • Output Markdown + HTML (base64 image embedding)

前置准备

Preparations

MinerU API Token

MinerU API Token

  1. 访问 https://mineru.net 注册账号
  2. 获取 API Token
  3. 设置环境变量（推荐）：
    bash
    export MINERU_TOKEN="your_token_here"
  1. Visit https://mineru.net to register an account
  2. Obtain the API Token
  3. Set environment variables (recommended):
    bash
    export MINERU_TOKEN="your_token_here"

依赖安装

Dependency Installation

bash
pip install requests markdown
bash
pip install requests markdown

操作步骤

Operation Steps

第一步：PDF 解析（使用 MinerU API）

Step 1: PDF Parsing (Using MinerU API)

bash
python scripts/mineru_api.py <pdf_path> <output_dir>
或者直接传入 token：
bash
python scripts/mineru_api.py paper.pdf ./output YOUR_TOKEN
输出结果：
  • output_dir/*.md
    - Markdown 文件（含公式、表格）
  • output_dir/images/
    - 高质量提取的图片
bash
python scripts/mineru_api.py <pdf_path> <output_dir>
Or directly pass the token:
bash
python scripts/mineru_api.py paper.pdf ./output YOUR_TOKEN
Output Results:
  • output_dir/*.md
    - Markdown files (including formulas, tables)
  • output_dir/images/
    - High-quality extracted images

第二步：提取论文信息

Step 2: Extract Paper Information

bash
python scripts/extract_paper_info.py <output_dir>/*.md paper_info.json
bash
python scripts/extract_paper_info.py <output_dir>/*.md paper_info.json

第三步：风格选择（询问用户）

Step 3: Style Selection (Ask User)

在生成文章前，必须询问用户以下选项：
Before generating the article, you must ask the user the following options:

1. 写作风格（必选）

1. Writing Style (Required)

风格特点适用场景
storytelling（故事型）从直觉出发，用比喻和例子，像讲故事公众号、技术博客、科普
academic（学术型）专业术语，严谨表述，保留原文概念学术报告、论文综述、研究组分享
concise（精炼型）直击核心，表格列表，信息密度高快速了解、论文速览、技术调研
StyleFeaturesApplication Scenarios
storytelling (Narrative)Start from intuition, use metaphors and examples like storytellingOfficial accounts, technical blogs, popular science
academic (Academic)Professional terminology, rigorous expression, retain original conceptsAcademic reports, paper reviews, research group sharing
concise (Concise)Get straight to the core, use tables and lists, high information densityQuick understanding, paper overview, technical research

2. 公式选项（可选）

2. Formula Options (Optional)

选项说明
with-formulas插入公式图片并详细讲解符号含义
no-formulas（默认）纯文字描述，不包含公式图片
OptionDescription
with-formulasInsert formula images and explain the meaning of symbols in detail
no-formulas (Default)Text-only description, no formula images included

3. 代码选项（可选，仅当论文有 GitHub 时）

3. Code Options (Optional, only if the paper has GitHub links)

选项说明
with-code克隆仓库，贴关键源码，代码与论文对照讲解
no-code（默认）不包含代码分析
询问示例：
请选择文章风格：
  1. academic - 学术型，专业严谨（默认推荐）
  2. storytelling - 故事型，朴素接地气
  3. concise - 精炼型，快速阅读
是否需要公式讲解？（论文包含数学公式时推荐） 是否需要结合 GitHub 代码分析？（检测到开源仓库：xxx）
如果用户不确定选哪个，默认使用 academic（学术型）风格。
OptionDescription
with-codeClone the repository, paste key source code, and explain by comparing code with the paper
no-code (Default)No code analysis included
Example of Inquiry:
Please select an article style:
  1. academic - Academic, professional and rigorous (recommended default)
  2. storytelling - Narrative, down-to-earth
  3. concise - Concise, quick reading
Do you need formula explanations? (Recommended if the paper contains mathematical formulas) Do you need to combine with GitHub code analysis? (Open-source repository detected: xxx)
If users are unsure, use the academic style by default.

第四步：智能生成文章

Step 4: Intelligent Article Generation

根据用户选择的风格，阅读对应的风格定义文件：
  • styles/storytelling.md
    - 故事型风格指南
  • styles/academic.md
    - 学术型风格指南
  • styles/concise.md
    - 精炼型风格指南
  • styles/with-formulas.md
    - 公式讲解指南
  • styles/with-code.md
    - 代码分析指南
Read the corresponding style definition file according to the user's selected style:
  • styles/storytelling.md
    - Storytelling Style Guide
  • styles/academic.md
    - Academic Style Guide
  • styles/concise.md
    - Concise Style Guide
  • styles/with-formulas.md
    - Formula Explanation Guide
  • styles/with-code.md
    - Code Analysis Guide

轻量模式（节省上下文）

Lightweight Mode (Save Context)

重要：为避免上下文膨胀，请遵循以下原则：
  1. 不要反复读取图片文件 - MinerU 已提取高质量图片，直接引用路径即可
  2. 信任 paper_info.json - 包含图片列表和元数据，无需视觉确认
  3. 只看关键图 - 最多读取 1-2 张核心架构图，其余直接引用
  4. 让用户验证 - 生成 HTML 后让用户自己检查图片是否正确
Important: To avoid context bloat, please follow these principles:
  1. Do not repeatedly read image files - MinerU has extracted high-quality images, just reference the path directly
  2. Trust paper_info.json - Contains image lists and metadata, no visual confirmation needed
  3. Only view key images - Read at most 1-2 core architecture diagrams, reference the rest directly
  4. Let users verify - After generating HTML, let users check if the images are correct themselves

通用写作原则

General Writing Principles

避免：
  • AI 常用词（"深入探讨"、"至关重要"、"在...领域"）
  • 机械化章节标题
  • LaTeX 公式语法（如 
    $\mathcal{O}(1)$
    ）- 使用提取的公式图片
  • 平铺直叙的技术描述
采用：
  • 自然段落叙述
  • 充分利用 MinerU 提取的图片
  • 论文中的每张关键图都应该被讲解到
  • 公式截图比 LaTeX 语法更易读
Avoid:
  • Overused AI phrases (e.g., "in-depth discussion", "of vital importance", "in the field of...")
  • Mechanized section titles
  • LaTeX formula syntax (e.g., 
    $\mathcal{O}(1)$
    ) - Use extracted formula images instead
  • Flat, straightforward technical descriptions
Adopt:
  • Natural paragraph narration
  • Make full use of images extracted by MinerU
  • Every key figure in the paper should be explained
  • Formula screenshots are more readable than LaTeX syntax

storytelling 风格方法论（故事型专用）

Storytelling Style Methodology (Exclusive for Storytelling Style)

以下方法论仅在用户选择 storytelling 风格时应用：
1. 从直觉切入，不要直接讲技术
  • 错误："本文提出了一种基于哈希表的条件记忆模块"
  • 正确："你有没有想过，大模型其实是没有记忆功能的？"
2. 先讲历史背景，再讲创新
  • 介绍新技术前，先解释相关的旧技术
  • 让读者理解"为什么需要这个创新"
3. 用简单例子贯穿全文
  • 选一个简单的例子反复使用
  • 例如："中国的首都在北京"
4. 使用生动的比喻
  • "大炮打蚊子"、"查字典 vs 背字典"
  • 让抽象概念具象化
5. 逻辑递进，层层深入
  • 简单问题 → 复杂问题 → 解决方案
6. 提炼核心洞见
  • 用一句话总结，如"记忆归记忆，计算归计算"
The following methodology only applies when users select the storytelling style:
1. Start from intuition, don't talk about technology directly
  • Wrong: "This paper proposes a conditional memory module based on hash tables"
  • Correct: "Have you ever thought that large models actually don't have memory functions?"
2. Talk about historical background first, then introduce innovations
  • Before introducing new technologies, explain related old technologies first
  • Let readers understand "why this innovation is needed"
3. Use simple examples throughout the article
  • Choose a simple example and reuse it repeatedly
  • For example: "The capital of China is Beijing"
4. Use vivid metaphors
  • "Using a cannon to kill a mosquito", "Looking up in a dictionary vs memorizing a dictionary"
  • Make abstract concepts concrete
5. Logical progression, layer by layer in-depth
  • Simple problems → Complex problems → Solutions
6. Refine core insights
  • Summarize in one sentence, e.g., "Memory is memory, computation is computation"

文章结构

Article Structure

1. 论文信息
markdown
**论文标题**：xxx
**论文链接**：[arXiv](https://arxiv.org/abs/xxxx)
**作者团队**：xxx
2. 直觉引入（2-3段）
  • 从一个问题或场景开始
  • 让读者产生好奇心
  • 引出"为什么需要这个研究"
3. 背景知识（3-4段）
  • 解释相关的基础技术或历史方法
  • 用简单例子说明
  • 让读者理解现有方案的局限
4. 核心创新（4-5段）
  • 详细讲解论文的创新点
  • 每个创新点都要有图片支撑
  • 用比喻和例子让抽象概念具象化
  • 公式用图片展示，不用 LaTeX 语法
5. 实验验证（2-3段）
  • 关键的实验结果图表
  • 对比分析和数据解读
  • 突出最亮眼的结果
6. 深入分析（2-3段）
  • 机制分析、消融实验等
  • 解释"为什么这个方法有效"
  • 提供更深层次的理解
7. 思考与展望（1-2段）
  • 提炼核心洞见
  • 预测未来发展方向
  • 个人观点和评价
1. Paper Information
markdown
**Paper Title**: xxx
**Paper Link**: [arXiv](https://arxiv.org/abs/xxxx)
**Author Team**: xxx
2. Intuition Introduction (2-3 paragraphs)
  • Start with a question or scenario
  • Arouse readers' curiosity
  • Lead to "why this research is needed"
3. Background Knowledge (3-4 paragraphs)
  • Explain relevant basic technologies or historical methods
  • Illustrate with simple examples
  • Let readers understand the limitations of existing solutions
4. Core Innovations (4-5 paragraphs)
  • Explain the paper's innovations in detail
  • Each innovation should be supported by images
  • Use metaphors and examples to make abstract concepts concrete
  • Display formulas with images instead of LaTeX syntax
5. Experimental Verification (2-3 paragraphs)
  • Key experimental result charts
  • Comparative analysis and data interpretation
  • Highlight the most outstanding results
6. In-depth Analysis (2-3 paragraphs)
  • Mechanism analysis, ablation experiments, etc.
  • Explain "why this method works"
  • Provide a deeper understanding
7. Thoughts and Outlook (1-2 paragraphs)
  • Refine core insights
  • Predict future development directions
  • Personal opinions and evaluations

第五步：输出格式（询问用户）

Step 5: Output Format (Ask User)

默认输出 Markdown，文章写完后询问用户是否需要其他格式：
"文章已生成：
article.md
。需要生成 HTML 版本吗？（HTML 会嵌入图片，方便直接分享）"
格式对比：
格式优势适用场景
MD（默认）轻量、易编辑、公众号可直接导入日常使用
HTML图片嵌入、单文件分享预览效果、分享给他人
如果用户需要 HTML：
bash
python scripts/generate_html.py <article.md> <output.html>
Markdown is the default output. After writing the article, ask users if they need other formats:
"Article generated: 
article.md
. Do you need to generate an HTML version? (HTML embeds images for easy direct sharing)"
Format Comparison:
FormatAdvantagesApplication Scenarios
MD (Default)Lightweight, easy to edit, can be directly imported into official accountsDaily use
HTMLImage embedding, single-file sharingPreview effects, sharing with others
If users need HTML:
bash
python scripts/generate_html.py <article.md> <output.html>

资源索引

Resource Index

风格定义：
  • styles/storytelling.md
    - 故事型风格
  • styles/academic.md
    - 学术型风格
  • styles/concise.md
    - 精炼型风格
  • styles/with-formulas.md
    - 公式讲解
  • styles/with-code.md
    - 代码分析
脚本：
  • scripts/mineru_api.py
    - MinerU Cloud API 调用（推荐）
  • scripts/convert_pdf.py
    - 本地转换（备选，需要 PyMuPDF）
  • scripts/extract_paper_info.py
    - 提取论文元数据
  • scripts/generate_html.py
    - 生成 HTML（base64 图片）
Style Definitions:
  • styles/storytelling.md
    - Storytelling Style
  • styles/academic.md
    - Academic Style
  • styles/concise.md
    - Concise Style
  • styles/with-formulas.md
    - Formula Explanations
  • styles/with-code.md
    - Code Analysis
Scripts:
  • scripts/mineru_api.py
    - MinerU Cloud API Call (Recommended)
  • scripts/convert_pdf.py
    - Local Conversion (Alternative, requires PyMuPDF)
  • scripts/extract_paper_info.py
    - Extract Paper Metadata
  • scripts/generate_html.py
    - Generate HTML (base64 images)

注意事项

Notes

  • 优先使用 MinerU API，精度最高，支持公式/表格
  • 节省上下文：不要反复读取图片，信任元数据
  • 不输出分析过程，用户只看最终文章
  • 避免分点列表，使用自然段落叙述
  • 图片选择 3-5 张关键图表
  • Prioritize using MinerU API for the highest accuracy, supporting formulas/tables
  • Save context: Do not repeatedly read images, trust metadata
  • Do not output the analysis process; users only need the final article
  • Avoid bullet points, use natural paragraph narration
  • Select 3-5 key charts as images

API 限制

API Limitations

  • 单个文件最大 200MB
  • 单个文件最多 600 页
  • 支持 PDF、DOC、PPT、图片等格式
  • Maximum size per file: 200MB
  • Maximum pages per file: 600 pages
  • Supports formats such as PDF, DOC, PPT, images, etc.