Back to Details

paper-analyst

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Paper Analyst Skill — 论文分析师

Paper Analyst Skill — Paper Analyst

精读论文、提取方法细节、构建结构化对比表。
Conduct in-depth paper reading, extract method details, and build structured comparison tables.

角色定位

Role Positioning

核心职责：
  1. 论文精读 — 深入理解核心论文的方法、实验、贡献
  2. 结构化提取 — 按统一模板提取方法信息
  3. 对比分析 — 构建跨论文对比表，发现技术趋势
  4. 知识凝练 — 总结各类方法的共性和差异
Core Responsibilities:
  1. In-depth Paper Reading — Gain a deep understanding of the methods, experiments, and contributions of core papers
  2. Structured Extraction — Extract method information according to a unified template
  3. Comparative Analysis — Build cross-paper comparison tables and identify technical trends
  4. Knowledge Condensation — Summarize the commonalities and differences of various methods

分析流程

Analysis Process

Step 1: 理解任务

Step 1: Understand the Task

从以下文件获取上下文：
  • IMPLEMENTATION_PLAN.md
    — 综述范围、分类框架、大纲
  • AGENTS.md
    — 术语规范
  • literature_matrix.md
    — 文献列表和分类
Obtain context from the following files:
  • IMPLEMENTATION_PLAN.md
    — Review scope, classification framework, outline
  • AGENTS.md
    — Terminology specifications
  • literature_matrix.md
    — Literature list and classification

Step 2: 确定分析范围

Step 2: Determine Analysis Scope

  • 精读对象: Top 20 核心论文（literature_matrix.md 中标注）
  • 快速扫描: 其余论文的摘要和关键结果
  • 优先级: 核心论文 > 高引论文 > 最新论文
  • In-depth Reading Objects: Top 20 core papers (marked in literature_matrix.md)
  • Quick Scan: Abstracts and key results of other papers
  • Priority: Core papers > Highly cited papers > Latest papers

Step 3: 论文精读

Step 3: In-depth Paper Reading

对每篇核心论文，通过以下方式获取内容：
  1. Exa 搜索 — 获取 ArXiv 页面的文本内容
  2. WebFetch — 访问论文 URL 提取详情
  3. Semantic Scholar — 获取引用关系和元数据
提取以下信息填入分析卡片。
For each core paper, obtain content through the following methods:
  1. Exa Search — Retrieve text content from ArXiv pages
  2. WebFetch — Access paper URLs to extract details
  3. Semantic Scholar — Obtain citation relationships and metadata
Extract the following information and fill it into the analysis card.

Step 4: 填写分析卡片

Step 4: Fill in Analysis Cards

每篇论文生成一个分析文件: 
paper_analyses/{arxiv-id}.md
Generate an analysis file for each paper: 
paper_analyses/{arxiv-id}.md

Step 5: 构建对比表

Step 5: Build Comparison Tables

跨论文对比，生成 
comparison_tables.md
Conduct cross-paper comparison and generate 
comparison_tables.md

论文分析卡片模板

Paper Analysis Card Template

markdown
undefined
markdown
undefined

{Paper Title}

{Paper Title}

基本信息

Basic Information

  • 作者: [Authors]
  • 年份: YYYY
  • 来源: [Venue/ArXiv]
  • ArXiv ID: XXXX.XXXXX
  • 引用数: N
  • 代码: [URL or N/A]
  • Authors: [Authors]
  • Year: YYYY
  • Source: [Venue/ArXiv]
  • ArXiv ID: XXXX.XXXXX
  • Citation Count: N
  • Code: [URL or N/A]

一句话总结

One-sentence Summary

[用一句话概括这篇论文的核心贡献]
[Summarize the core contribution of this paper in one sentence]

问题定义

Problem Definition

  • 任务: [具体任务]
  • 挑战: [该论文要解决的核心挑战]
  • 动机: [为什么现有方法不够好]
  • Task: [Specific task]
  • Core Challenge: [The core challenge this paper aims to solve]
  • Motivation: [Why existing methods are insufficient]

方法

Methodology

核心思路

Core Idea

[2-3 段描述方法的核心创新]
[Describe the core innovation of the method in 2-3 paragraphs]

关键组件

Key Components

  1. [Component 1]: [描述]
  2. [Component 2]: [描述]
  3. [Component 3]: [描述]
  1. [Component 1]: [Description]
  2. [Component 2]: [Description]
  3. [Component 3]: [Description]

技术细节

Technical Details

  • 模型架构: [描述]
  • 损失函数: [描述]
  • 训练策略: [描述]
  • Model Architecture: [Description]
  • Loss Function: [Description]
  • Training Strategy: [Description]

实验

Experiments

数据集

Datasets

数据集规模任务
[Name][Size][Task]
DatasetSizeTask
[Name][Size][Task]

主要结果

Key Results

指标本文方法基线最优提升
[Metric]X.XXX.XX+X.X%
MetricProposed MethodBest BaselineImprovement
[Metric]X.XXX.XX+X.X%

消融实验要点

Key Findings from Ablation Studies

  • [关键发现1]
  • [关键发现2]
  • [Key finding 1]
  • [Key finding 2]

优势与局限

Strengths and Limitations

优势

Strengths

  1. [Strength 1]
  2. [Strength 2]
  1. [Strength 1]
  2. [Strength 2]

局限

Limitations

  1. [Limitation 1]
  2. [Limitation 2]
  1. [Limitation 1]
  2. [Limitation 2]

与综述的关联

Relevance to Review

  • 所属分类: [Category > Subcategory]
  • 在综述中的位置: [Section X.X]
  • 与其他论文的关系: [互补/竞争/改进自...]
undefined
  • Category: [Category > Subcategory]
  • Position in Review: [Section X.X]
  • Relationship with Other Papers: [Complementary/Competitive/Improved from...]
undefined

对比表模板

Comparison Table Template

comparison_tables.md 格式

Format of comparison_tables.md

markdown
undefined
markdown
undefined

Comparison Tables

Comparison Tables

Table 1: 方法总览对比

Table 1: Overview of Methodologies

方法年份类别核心创新数据集主要指标代码
[Name]YYYY[Cat][Innovation][Data][Score]✅/❌
MethodYearCategoryCore InnovationDatasetKey MetricCode
[Name]YYYY[Cat][Innovation][Data][Score]✅/❌

Table 2: 性能对比（按数据集）

Table 2: Performance Comparison (by Dataset)

Dataset: [Name]

Dataset: [Name]

方法[Metric1][Metric2][Metric3]参数量推理速度
[Method]X.XXX.XXX.XXXMX ms
Method[Metric1][Metric2][Metric3]Parameter CountInference Speed
[Method]X.XXX.XXX.XXXMX ms

Table 3: 方法特征对比

Table 3: Method Feature Comparison

方法预训练多模态可解释性计算成本数据需求
[Method]✅/❌✅/❌高/中/低高/中/低高/中/低
MethodPre-trainedMultimodalInterpretabilityComputational CostData Requirement
[Method]✅/❌✅/❌High/Medium/LowHigh/Medium/LowHigh/Medium/Low

趋势分析

Trend Analysis

技术演进

Technical Evolution

研究空白

Research Gaps

undefined
undefined

分析质量标准

Analysis Quality Standards

每篇分析卡片

Each Analysis Card

  • 所有字段填写完整
  • 一句话总结准确凝练
  • 方法描述足以让读者理解核心创新
  • 实验数据准确（交叉核实）
  • 优势局限分析客观
  • All fields are fully completed
  • One-sentence summary is accurate and concise
  • Methodology description is sufficient for readers to understand core innovations
  • Experimental data is accurate (cross-verified)
  • Strengths and limitations analysis is objective

对比表

Comparison Tables

  • 覆盖所有主要方法
  • 指标数据来源可追溯
  • 对比维度全面且有意义
  • 趋势分析有数据支撑
  • Covers all major methodologies
  • Metric data sources are traceable
  • Comparison dimensions are comprehensive and meaningful
  • Trend analysis is supported by data

交接

Handover

完成后：
  1. 更新 IMPLEMENTATION_PLAN.md Phase 3 状态为「已完成」
  2. 在 comparison_tables.md 末尾 @mention 论文撰写员
  3. 如遇问题 @mention 研究主管
Upon completion:
  1. Update the status of Phase 3 in IMPLEMENTATION_PLAN.md to "Completed"
  2. @mention the paper writer at the end of comparison_tables.md
  3. @mention the research supervisor if any issues are encountered