learning-reflector

name	Learning Reflector
description	"深度反思学习过程，巩固知识，发现盲点，优化学习策略。基于笔记内容生成反思报告、学习建议、知识图谱。"

When to Use

Use this skill when you want to:

• Deeply reflect on what you learned
• Consolidate knowledge and find blind spots
• Get personalized learning suggestions
• Discover connections between concepts
• Optimize your learning strategy

Triggers:

• User says "reflect on learning", "反思学习", "学习反思"
• User mentions "knowledge gaps", "知识盲点"
• User asks "what should I learn next", "接下来学什么"
• User wants "learning suggestions", "学习建议"
• User mentions "consolidate knowledge", "巩固知识"

Quick Start

Reflect on Today's Learning

反思今天的学习内容

Weekly Reflection

生成本周学习反思报告

Get Learning Suggestions

分析我的学习盲点并给出建议

Manual Run

python {workspace}/PaperVault/scripts/learning-reflect.py

Features

1. Deep Reflection (深度反思)

Analyzes your notes to generate insights:

• Key concepts learned
• Understanding depth (surface vs deep)
• Connections between papers
• Knowledge gaps identified
• Questions to explore

2. Knowledge Consolidation (知识巩固)

Helps you remember and understand:

• Spaced repetition reminders
• Concept maps generation
• Analogy creation
• Example applications

3. Blind Spot Discovery (盲点发现)

Identifies what you might have missed:

• Incomplete understanding
• Related topics not explored
• Foundational concepts to review
• Contradicting viewpoints

4. Learning Optimization (学习优化)

Personalized suggestions:

• Reading priority recommendations
• Time allocation suggestions
• Learning path optimization
• Resource recommendations

Reflection Report Structure

---
type: reflection
date: YYYY-MM-DD
period: daily|weekly|monthly
---

# 🧠 学习反思报告 - YYYY-MM-DD

## 学习概览

- **论文数量**: X 篇
- **核心概念**: X 个
- **理解深度**: ⭐⭐⭐☆☆ (3/5)

## 核心收获

### 1. 概念理解
- **Transformer 架构**: 
  - 理解了自注意力机制
  - 掌握了位置编码原理
  - 需要深入: 多头注意力的细节

### 2. 方法学习
- **RAG 优化**:
  - 学习了检索增强策略
  - 了解了重排序方法
  - 待探索: 混合检索方案

## 知识连接

```mermaid
graph TD
    A[Transformer] --> B[Attention]
    B --> C[Self-Attention]
    B --> D[Multi-Head]
    A --> E[Position Encoding]
    F[RAG] --> G[Retrieval]
    F --> H[Generation]
    C --> H

盲点发现

❓ 理解不完整

1. Layer Normalization: 了解了作用，但不清楚与 Batch Norm 的区别
2. Beam Search: 听说过，但没有深入理解

🔍 相关但未探索

1. Efficient Transformers: 与标准 Transformer 的对比
2. Vision Transformers: 在图像领域的应用

📚 基础需加强

1. 概率图模型: 是理解某些论文的基础
2. 强化学习: Agent 论文中经常用到

巩固建议

今日复习

• 重读 Transformer 论文第 3 节
• 手推一次注意力公式
• 画出完整的架构图

本周任务

• 对比 3 种位置编码方法
• 实现一个简单的 RAG demo
• 阅读补充材料

学习建议

🎯 优先级调整

1. 高优先级: 深入理解 Transformer 细节
2. 中优先级: RAG 优化方法实践
3. 低优先级: 扩展阅读其他架构

⏰ 时间分配

• 理论: 40% (深入理解原理)
• 实践: 40% (代码实现)
• 阅读: 20% (扩展视野)

📖 推荐资源

1. Transformer 可视化: https://jalammar.github.io/illustrated-transformer/
2. RAG 实战教程: LangChain 官方文档
3. 论文: "Attention Is All You Need" (再读一遍)

下一步计划

即时行动

1. 完成今日复习任务
2. 整理 Transformer 笔记
3. 准备 RAG 实践环境

本周目标

1. 完成 Transformer 深度学习
2. 实现第一个 RAG demo
3. 阅读 3 篇相关论文

长期规划

1. 建立 Transformer 知识体系
2. 掌握 RAG 最佳实践
3. 探索 Agent 应用场景

自我评估

维度	评分	说明
理解深度	⭐⭐⭐☆☆	基本概念清晰，细节需加强
知识连接	⭐⭐⭐⭐☆	能发现概念间联系
实践能力	⭐⭐☆☆☆	理论多，实践少
持续性	⭐⭐⭐⭐☆	坚持每日学习

反思日记

今天学习了 Transformer 和 RAG 的基础知识。虽然理解了主要概念，但感觉深度不够，特别是数学推导部分。明天需要花更多时间在公式理解上，而不是只看文字描述。

实践方面明显不足，需要找时间动手实现。计划本周完成一个简单的 RAG demo，这样能更好地理解理论。

整体来说，学习方向正确，但需要调整理论和实践的比例。

## Reflection Types

### 1. Quick Reflection (5 minutes)
**When**: After reading each paper
**Focus**: 
- Main takeaway (1 sentence)
- Questions (1-2)
- Next steps (1-2)

### 2. Daily Reflection (15 minutes)
**When**: End of each day
**Focus**:
- What I learned today
- What confused me
- What to review tomorrow

### 3. Weekly Reflection (30 minutes)
**When**: End of each week
**Focus**:
- Weekly themes
- Knowledge connections
- Progress towards goals

### 4. Monthly Reflection (1 hour)
**When**: End of each month
**Focus**:
- Monthly achievements
- Knowledge map evolution
- Strategy adjustment

## Prompt Templates

### Concept Understanding Check

基于我的笔记，请帮我反思对 "{concept}" 的理解：

1. 我理解的层次:

   • 能复述定义
   • 能解释原理
   • 能举例说明
   • 能发现联系
   • 能批判评价

2. 可能的理解盲点:

3. 建议深入的方向:

### Knowledge Connection Discovery

请分析以下概念之间的联系： {concept1}, {concept2}, {concept3}

1. 直接联系:
2. 间接联系:
3. 共同基础:
4. 应用场景重叠:

### Learning Strategy Optimization

基于我过去 {period} 的学习记录：

1. 时间分配分析:

   • 阅读时间: X%
   • 思考时间: X%
   • 实践时间: X%

2. 效率评估:

   • 高效时段:
   • 低效时段:

3. 优化建议:

## Advanced Features

### Spaced Repetition
Automatically suggests concepts to review based on:
- Ebbinghaus forgetting curve
- Your understanding depth
- Concept importance

### Knowledge Graph Generation
Creates visual maps of:
- Concept relationships
- Learning paths
- Knowledge gaps

### Comparative Analysis
Compares your learning with:
- Previous periods
- Recommended paths
- Expert roadmaps

## Integration with Other Skills

- **paper-fetcher**: Source of papers to reflect on
- **paper-summarizer**: Summaries to analyze
- **pdf-reader**: Deep paper understanding
- **self-improving**: Long-term memory

## Troubleshooting

### Not Enough Data

Error: Insufficient notes for reflection

**Solution**: Use paper-fetcher to get more papers first.

### Superficial Reflection

Warning: Reflection too shallow

**Solution**: 
1. Add more detailed notes
2. Use pdf-reader for deeper analysis
3. Increase reflection time

## Best Practices

1. **Reflect Daily**: 15 minutes at end of day
2. **Be Honest**: Admit what you don't understand
3. **Take Action**: Act on the suggestions
4. **Track Progress**: Review reflections over time
5. **Stay Curious**: Follow the questions that arise

## Related Skills

- `paper-fetcher` - Paper retrieval
- `paper-summarizer` - Progress tracking
- `pdf-reader` - Deep understanding
- `self-improving` - Long-term memory