graph-thinking
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseGraph Thinking - Non-Linear Problem Solving
图思维——非线性问题解决
Mental model for visualizing complex relationships and connections between
ideas, concepts, or data points. Evolved from Graph-of-Thought (GoT) reasoning
that mirrors human cognition.
这是一种用于可视化想法、概念或数据点之间复杂关系与连接的思维模型,由模仿人类认知的Graph-of-Thought(GoT,图思维推理)发展而来。
When to Use This Skill
何时使用此技能
- Mapping feature dependencies in product development
- Analyzing stakeholder relationships
- Understanding system architectures
- Exploring interconnected concepts
- Designing recommendation systems or knowledge graphs
- Identifying opportunity areas through network analysis
- 产品开发中映射功能依赖关系
- 分析利益相关者关系
- 理解系统架构
- 探索相互关联的概念
- 设计推荐系统或知识图谱
- 通过网络分析识别机会领域
Core Concepts
核心概念
Graph Elements
图元素
| Element | Description |
|---|---|
| Nodes | Individual elements or concepts |
| Edges | Relationships or connections between nodes |
| Clusters | Groups of highly connected nodes |
| Pathways | Routes through the network |
| Centrality | Measures identifying most important nodes |
| Topology | Structural arrangement of connections |
| 元素 | 描述 |
|---|---|
| Nodes(节点) | 单个元素或概念 |
| Edges(边) | 节点之间的关系或连接 |
| Clusters(集群) | 高度连接的节点组 |
| Pathways(路径) | 网络中的传播路线 |
| Centrality(中心性) | 识别最重要节点的衡量指标 |
| Topology(拓扑) | 连接的结构排列 |
Graph-of-Thought (GoT) Reasoning
Graph-of-Thought (GoT) 推理
Traditional (Chain-of-Thought):
A → B → C → D → Conclusion
Graph-of-Thought:
┌─── B ───┐
│ │
A ──┼─── C ───┼──→ Synthesis → Conclusion
│ │
└─── D ───┘
↑
Feedback LoopGoT enables:
- Combining arbitrary thoughts into synergistic outcomes
- Distilling networks of thoughts for clarity
- Enhancing ideas using feedback loops
- Non-linear exploration of solution spaces
Traditional (Chain-of-Thought):
A → B → C → D → Conclusion
Graph-of-Thought:
┌─── B ───┐
│ │
A ──┼─── C ───┼──→ Synthesis → Conclusion
│ │
└─── D ───┘
↑
Feedback LoopGoT 支持:
- 将任意想法组合成协同成果
- 提炼思维网络以提升清晰度
- 通过反馈环优化想法
- 非线性探索解决方案空间
Fundamental Principles
基本原则
First Principles Thinking
第一性原理思维
Break down complex problems into fundamental truths:
Surface Level:
"We need more marketing"
↓
Why?
↓
"Not enough customers"
↓
Why?
↓
Root Truth:
"Value proposition unclear to target audience"将复杂问题拆解为基本事实:
表面层级:
"我们需要更多营销"
↓
为什么?
↓
"客户数量不足"
↓
为什么?
↓
核心事实:
"目标受众对价值主张理解不清晰"Second-Order Thinking
二阶思维
Demand deeper analysis by asking "And then what?":
Decision: Reduce prices by 20%
First-order: More sales
Second-order: Lower margins → Less R&D budget
Third-order: Competitors catch up → Price war
Fourth-order: Race to bottom → Industry commoditization通过提问“然后呢?”进行深度分析:
决策:降价20%
一阶影响: 销量增加
二阶影响: 利润率降低 → 研发预算减少
三阶影响: 竞争对手跟进 → 价格战
四阶影响: 陷入恶性竞争 → 行业同质化Non-Linear Processing
非线性处理
Unlike sequential thinking:
| Sequential | Graph-Based |
|---|---|
| One path at a time | Multiple paths simultaneously |
| Linear information flow | Multi-directional exploration |
| Fixed order | Iterative refinement through loops |
| Single conclusion | Synthesized insights from multiple angles |
与顺序思维的对比:
| 顺序思维 | 基于图的思维 |
|---|---|
| 一次仅走一条路径 | 同时探索多条路径 |
| 线性信息流 | 多方向探索 |
| 固定顺序 | 通过循环迭代优化 |
| 单一结论 | 从多角度综合洞察 |
Analysis Framework
分析框架
Double Diamond Model
双钻石模型
Apply divergent and convergent thinking cycles:
DISCOVER DEFINE DEVELOP DELIVER
(Diverge) (Converge) (Diverge) (Converge)
/\ \/ /\ \/
/ \ / \ / \ / \
/ \ / \ / \ / \
/ \ / \ / \ / \
/ \ / \ / \ / \
Explore Focus on Generate Focus on
problem specific diverse optimal
space challenges solutions implementation应用发散与收敛思维循环:
DISCOVER(探索) DEFINE(定义) DEVELOP(开发) DELIVER(交付)
(发散) (收敛) (发散) (收敛)
/\ \/ /\ \/
/ \ / \ / \ / \
/ \ / \ / \ / \
/ \ / \ / \ / \
/ \ / \ / \ / \
探索 聚焦于 生成 聚焦于
问题空间 特定挑战 多样化解决方案 最优
实施方案Step 1: Map the Nodes
步骤1:映射节点
Identify all relevant elements:
Product Launch Analysis:
Nodes:
├── Stakeholders
│ ├── Customers
│ ├── Engineering
│ ├── Marketing
│ └── Leadership
├── Features
│ ├── Core functionality
│ ├── Nice-to-haves
│ └── Technical debt
├── Constraints
│ ├── Timeline
│ ├── Budget
│ └── Resources
└── Dependencies
├── External APIs
├── Infrastructure
└── Regulatory识别所有相关元素:
产品发布分析:
节点:
├── 利益相关者
│ ├── 客户
│ ├── 工程团队
│ ├── 营销团队
│ └── 管理层
├── 功能
│ ├── 核心功能
│ ├── 锦上添花的功能
│ └── 技术债务
├── 约束条件
│ ├── 时间线
│ ├── 预算
│ └── 资源
└── 依赖关系
├── 外部API
├── 基础设施
└── 合规要求Step 2: Define Relationships (Edges)
步骤2:定义关系(边)
Document connections between nodes:
Edge Types:
├── Dependency: A requires B
├── Influence: A affects B
├── Correlation: A and B move together
├── Conflict: A competes with B
└── Synergy: A enhances B记录节点之间的连接:
边类型:
├── 依赖:A需要B
├── 影响:A作用于B
├── 关联:A与B同步变化
├── 冲突:A与B竞争
└── 协同:A增强BStep 3: Identify Clusters and Patterns
步骤3:识别集群与模式
Find highly connected groups:
High Centrality (Critical Nodes):
├── Authentication service → 12 dependencies
├── Database layer → 8 dependencies
└── API gateway → 6 dependencies
Clusters:
├── User-facing features (tightly coupled)
├── Backend services (loosely coupled)
└── Third-party integrations (isolated)找到高度连接的组:
高中心性(关键节点):
├── 认证服务 → 12个依赖项
├── 数据库层 → 8个依赖项
└── API网关 → 6个依赖项
集群:
├── 用户端功能(紧密耦合)
├── 后端服务(松散耦合)
└── 第三方集成(独立)Step 4: Analyze Pathways
步骤4:分析路径
Trace routes through the network:
User Journey Graph:
Landing Page
↓
[Sign Up] ←→ [Social Login]
↓
Onboarding
↓ ↓
Quick Start Full Setup
↓ ↓
└─────┬─────┘
↓
First Value
↓
↙ ↓ ↘
Churn Retain Upgrade追踪网络中的传播路线:
用户旅程图:
着陆页
↓
[注册] ←→ [社交登录]
↓
引导流程
↓ ↓
快速开始 完整设置
↓ ↓
└─────┬─────┘
↓
首次价值体验
↓
↙ ↓ ↘
流失 留存 升级Output Template
输出模板
After completing analysis, document as:
markdown
undefined完成分析后,按以下格式记录:
markdown
undefinedGraph Thinking Analysis
图思维分析
Subject: [What you're analyzing]
Analysis Date: [Date]
主题: [分析对象]
分析日期: [日期]
Node Map
节点映射
| Category | Nodes | Centrality |
|---|---|---|
| [Cat 1] | [Nodes] | [High/Med/Low] |
| [Cat 2] | [Nodes] | [High/Med/Low] |
| 类别 | 节点 | 中心性 |
|---|---|---|
| [类别1] | [节点] | [高/中/低] |
| [类别2] | [节点] | [高/中/低] |
Relationship Matrix
关系矩阵
| From | To | Relationship | Strength |
|---|---|---|---|
| [A] | [B] | [Type] | [1-5] |
| 来源 | 目标 | 关系类型 | 强度 |
|---|---|---|---|
| [A] | [B] | [类型] | [1-5] |
Key Insights
关键洞察
- Clusters identified: [Description]
- Critical paths: [Description]
- Bottlenecks: [Description]
- Opportunities: [Description]
- 识别的集群: [描述]
- 关键路径: [描述]
- 瓶颈: [描述]
- 机会: [描述]
Recommendations
建议
| Priority | Action | Rationale |
|---|---|---|
| High | [Action] | [Why] |
| Medium | [Action] | [Why] |
undefined| 优先级 | 行动 | 理由 |
|---|---|---|
| 高 | [行动] | [原因] |
| 中 | [行动] | [原因] |
undefinedApplication Examples
应用示例
Feature Dependency Mapping
功能依赖映射
Feature: Real-time Collaboration
Dependencies:
├── WebSocket infrastructure
│ ├── Connection management
│ └── Message queuing
├── Conflict resolution
│ ├── Operational transforms
│ └── CRDT implementation
├── Presence indicators
│ └── User state sync
└── Permissions
├── Document access
└── Cursor visibility功能:实时协作
依赖项:
├── WebSocket基础设施
│ ├── 连接管理
│ └── 消息队列
├── 冲突解决
│ ├── 操作转换
│ └── CRDT实现
├── 在线状态指示器
│ └── 用户状态同步
└── 权限
├── 文档访问
└── 光标可见性Stakeholder Analysis
利益相关者分析
HIGH INFLUENCE
│
Keep Satisfied │ Manage Closely
┌─────────────────────┼─────────────────────┐
│ │ │
│ Executives │ Product Owner │
│ Compliance │ Key Customers │
│ │ │
LOW ──────────────────────┼────────────────────── HIGH
INTEREST │ INTEREST
│ │ │
│ General Users │ Power Users │
│ IT Support │ Dev Team │
│ │ │
└─────────────────────┼─────────────────────┘
Monitor │ Keep Informed
│
LOW INFLUENCE 高影响力
│
保持满意 │ 密切管理
┌─────────────────────┼─────────────────────┐
│ │ │
│ 高管 │ 产品负责人 │
│ 合规团队 │ 核心客户 │
│ │ │
低 ──────────────────────┼────────────────────── 高
关注度 │ 关注度
│ │ │
│ 普通用户 │ 核心用户 │
│ IT支持团队 │ 开发团队 │
│ │ │
└─────────────────────┼─────────────────────┘
监控 │ 保持告知
│
低影响力System Architecture Analysis
系统架构分析
Microservice Graph:
API Gateway [Centrality: 0.95]
│
├── Auth Service [0.82]
│ └── User DB
│
├── Product Service [0.71]
│ ├── Catalog DB
│ └── Search Index
│
├── Order Service [0.68]
│ ├── Order DB
│ └── Payment Gateway (external)
│
└── Notification Service [0.45]
└── Email Provider (external)
Critical Path: Gateway → Auth → Product → Order
Bottleneck: Auth Service (single point of failure)微服务图:
API网关 [中心性: 0.95]
│
├── 认证服务 [0.82]
│ └── 用户数据库
│
├── 产品服务 [0.71]
│ ├── 目录数据库
│ └── 搜索索引
│
├── 订单服务 [0.68]
│ ├── 订单数据库
│ └── 支付网关(外部)
│
└── 通知服务 [0.45]
└── 邮件服务商(外部)
关键路径: 网关 → 认证 → 产品 → 订单
瓶颈: 认证服务(单点故障)Best Practices
最佳实践
Do
建议做法
- Visualize relationships - Draw the graph, don't just describe it
- Iterate continuously - Graphs evolve as understanding deepens
- Measure centrality - Identify the most critical nodes
- Look for clusters - Natural groupings reveal system structure
- Trace pathways - Understand how information/value flows
- 可视化关系 - 绘制图,而不只是描述
- 持续迭代 - 随着理解加深,图会不断演变
- 衡量中心性 - 识别最关键的节点
- 寻找集群 - 自然分组能揭示系统结构
- 追踪路径 - 理解信息/价值的流动方式
Avoid
避免事项
- Over-connecting - Not everything relates to everything
- Ignoring edge types - Different relationships have different meanings
- Static thinking - Graphs change over time
- Missing feedback loops - Circular dependencies are significant
- Forgetting weights - Some relationships are stronger than others
- 过度连接 - 并非所有事物都相互关联
- 忽略边类型 - 不同关系有不同含义
- 静态思维 - 图会随时间变化
- 遗漏反馈循环 - 循环依赖很重要
- 忘记权重 - 有些关系比其他关系更强
Integration with Other Methods
与其他方法的整合
| Method | Combined Use |
|---|---|
| Five Whys | Trace causal chains through the graph |
| Business Canvas | Map relationships between canvas elements |
| Jobs-to-be-Done | Connect user needs to feature nodes |
| Hypothesis Tree | Structure experiments as branching graphs |
| Stakeholder Map | Visualize influence and interest relationships |
| 方法 | 组合用途 |
|---|---|
| 五个为什么 | 通过图追踪因果链 |
| 商业模式画布 | 映射画布元素之间的关系 |
| Jobs-to-be-Done(用户待办任务) | 将用户需求与功能节点连接 |
| 假设树 | 将实验构建为分支图 |
| 利益相关者地图 | 可视化影响力与关注度关系 |
Tools
工具
Visualization
可视化
- Mermaid - Code-based diagrams in markdown
- Graphviz - Programmatic graph generation
- Excalidraw - Hand-drawn style diagrams
- Miro/FigJam - Collaborative whiteboarding
- Mermaid - Markdown中的代码驱动图表
- Graphviz - 程序化图生成工具
- Excalidraw - 手绘风格图表工具
- Miro/FigJam - 协作白板工具
Analysis
分析
- Gephi - Network analysis and visualization
- Neo4j - Graph database for complex queries
- NetworkX - Python library for graph algorithms
- Gephi - 网络分析与可视化工具
- Neo4j - 用于复杂查询的图数据库
- NetworkX - Python图算法库