fishbone-diagram

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Fishbone Diagram (Ishikawa) Analysis

鱼骨图（石川图）分析

Create structured cause-and-effect diagrams to systematically identify potential root causes of problems. This skill guides collaborative brainstorming, ensures comprehensive category coverage, and produces visual outputs.

创建结构化的因果图，系统识别问题的潜在根本原因。该方法引导协作式头脑风暴，确保全面覆盖类别，并生成可视化输出。

Integration with Other RCCA Tools

与其他RCCA工具的集成

The Fishbone Diagram provides breadth (identifying all possible causes across categories), while 5 Whys provides depth (drilling into specific causes). Typical workflow:

Use Fishbone to brainstorm and categorize all potential causes
Prioritize top 2-3 causes via multi-voting
Apply 5 Whys to each prioritized cause to find root causes

Also integrates with: Pareto Analysis (prioritize by frequency/impact), FMEA (risk assessment), 8D (Problem Definition phase).

鱼骨图提供广度（识别跨类别的所有可能原因），而5Why分析法提供深度（深挖特定原因）。典型工作流：

使用鱼骨图进行头脑风暴并分类所有潜在原因
通过多投票选出前2-3个优先级最高的原因
对每个优先级原因应用5Why分析法以找到根本原因

还可与以下工具集成：帕累托分析（按频率/影响排序）、FMEA（风险评估）、8D（问题定义阶段）。

Workflow Overview

工作流概述

6 Phases (Q&A-driven):

Problem Definition → Clear, specific effect statement
Category Selection → Choose framework (6Ms/8Ps/4Ss/custom)
Cause Brainstorming → Identify causes under each category
Sub-cause Drilling → Add 2-3 levels of detail
Prioritization → Multi-voting to identify top causes
Documentation → Generate diagram and report

6个阶段（问答驱动）：

问题定义 → 清晰、具体的效果陈述
类别选择 → 选择框架（6Ms/8Ps/4Ss/自定义）
原因头脑风暴 → 识别每个类别下的原因
子原因深挖 → 添加2-3层细节
优先级排序 → 多投票确定关键原因
文档记录 → 生成图表和报告

Phase 1: Problem Definition

阶段1：问题定义

Goal: Establish a clear, specific, measurable problem statement.

Ask the user:

What specific problem or effect are you trying to analyze?

A good problem statement is:

Specific: "Machine 4 overheated at 2 PM" not "Machine broke"

Measurable: Include quantities, frequencies, or timeframes when possible

Observable: Describes what happened, not why

Non-blaming: Focus on the situation, not individuals

Quality Gate: Problem statement must:

Describe observable effect (not assumed cause)
Be specific enough to guide focused analysis
Avoid embedding solutions or blame

If vague, ask: "Can you be more specific about [what/when/where/how much]?"

目标：建立清晰、具体、可衡量的问题陈述。

询问用户：

你试图分析的具体问题或结果是什么？

一个好的问题陈述应：

具体：例如“4号机器在下午2点过热”而非“机器坏了”

可衡量：尽可能包含数量、频率或时间范围

可观察：描述发生了什么，而非原因

无指责：聚焦于情况，而非个人

质量门：问题陈述必须：

描述可观察的结果（而非假设的原因）
足够具体以指导聚焦分析
避免嵌入解决方案或指责

如果模糊，询问：“你能更具体地说明[什么/何时/何地/程度]吗？”

Phase 2: Category Selection

阶段2：类别选择

Goal: Select appropriate cause categories for the analysis context.

Present options:

Which category framework fits your analysis context?

6Ms (Manufacturing/Operations):

Man (People), Machine, Method, Material, Measurement, Mother Nature (Environment)

8Ps (Service/Marketing):

Product, Price, Place, Promotion, People, Process, Physical Evidence, Policies

4Ss (Service Operations):

Surroundings, Suppliers, Systems, Skills

Custom: Define your own categories based on your specific domain

Or describe your context and I'll recommend an appropriate framework.

For detailed category definitions and prompting questions, see:

references/category-frameworks.md

目标：为分析场景选择合适的原因类别。

提供选项：

哪个类别框架适合你的分析场景？

6Ms（制造/运营）：

人员（Man）、机器（Machine）、方法（Method）、材料（Material）、测量（Measurement）、环境（Mother Nature）

8Ps（服务/营销）：

产品（Product）、价格（Price）、渠道（Place）、推广（Promotion）、人员（People）、流程（Process）、有形证据（Physical Evidence）、政策（Policies）

4Ss（服务运营）：

环境（Surroundings）、供应商（Suppliers）、系统（Systems）、技能（Skills）

自定义：根据你的特定领域定义自己的类别

或者描述你的场景，我会推荐合适的框架。

有关详细的类别定义和提示问题，请参阅：

references/category-frameworks.md

Phase 3: Cause Brainstorming

阶段3：原因头脑风暴

Goal: Generate comprehensive list of potential causes under each category.

For each category, ask:

Under [Category], what factors might contribute to "[Problem]"?

Think about:

What could go wrong in this area?

What variations or inconsistencies exist?

What has changed recently?

Facilitation techniques (see

references/facilitation-guide.md

Round-robin: Each participant contributes one cause, rotate until exhausted
Brainwriting: Silent individual brainstorming on sticky notes before discussion
Affinity grouping: Cluster related causes together
"Why does this happen?": Probe each cause for deeper understanding

Quality indicators:

Minimum 2-3 causes per category (empty categories may indicate blind spots)
Mix of obvious and non-obvious causes
Causes should be distinct (not restating the problem)

目标：生成每个类别下的潜在原因的完整列表。

针对每个类别，询问：

在**[类别]**下，哪些因素可能导致“[问题]”？

思考：

这个领域可能出现什么问题？

存在哪些差异或不一致？

最近有什么变化？

引导技巧（参阅

references/facilitation-guide.md

）：

循环发言：每位参与者提出一个原因，轮流直到无新想法
书面头脑风暴：先进行静默个人头脑风暴，将想法写在便利贴上，再讨论
亲和分组：将相关原因聚类
“为什么会发生？”：对每个原因进行深挖以获得更深入的理解

质量指标：

每个类别至少2-3个原因（空类别可能表示存在盲区）
混合明显和非明显的原因
原因应独特（而非重复问题）

Phase 4: Sub-cause Drilling

阶段4：子原因深挖

Goal: Add depth to major causes with 2-3 levels of sub-causes.

For significant causes, ask:

For the cause "[Cause]", what specific factors contribute to it?

Ask "Why might this happen?" to uncover sub-causes.

Depth guidance:

Level 1: Direct causes (e.g., "Equipment malfunction")
Level 2: Contributing factors (e.g., "Lack of maintenance")
Level 3: Root-level factors (e.g., "No maintenance schedule defined")

Typically 2-3 levels is sufficient. If more depth needed, transition to 5 Whys analysis.

目标：为主要原因添加2-3层子原因以增加深度。

针对重要原因，询问：

对于原因“[原因]”，哪些具体因素会导致它？

问“为什么会发生这种情况？”以发现子原因。

深度指南：

第1层：直接原因（例如“设备故障”）
第2层：促成因素（例如“缺乏维护”）
第3层：根级因素（例如“未定义维护计划”）

通常2-3层足够。如果需要更深入，可过渡到5Why分析。

Phase 5: Prioritization

阶段5：优先级排序

Goal: Identify most likely/impactful causes for focused investigation.

Present prioritization options:

How would you like to prioritize the identified causes?

Multi-voting (Recommended): Each participant gets 3 votes to place on causes they believe are most significant

Impact-Effort Matrix: Rate each cause by impact (if addressed) and effort (to investigate/fix)

Data-driven: Use existing data to identify most frequent/costly causes (Pareto)

Consensus: Team discussion to agree on top 3-5 causes

After prioritization:

The top prioritized causes are:

[Cause 1] - [votes/score]

[Cause 2] - [votes/score]

[Cause 3] - [votes/score]

Would you like to apply 5 Whys analysis to drill deeper into any of these?

目标：确定最可能/影响最大的原因，以便聚焦调查。

提供优先级排序选项：

你希望如何对已识别的原因进行优先级排序？

多投票（推荐）：每位参与者获得3票，投给他们认为最重要的原因

影响-努力矩阵：根据影响（解决后的效果）和努力（调查/修复的成本）对每个原因评分

数据驱动：使用现有数据识别最频繁/成本最高的原因（帕累托分析）

共识：团队讨论以达成一致的前3-5个原因

优先级排序后：

优先级最高的原因是：

[原因1] - [票数/分数]

[原因2] - [票数/分数]

[原因3] - [票数/分数]

你是否想对其中任何一个原因应用5Why分析法进行更深入的挖掘？

Phase 6: Documentation

阶段6：文档记录

Goal: Generate visual diagram and comprehensive report.

Ask:

Ready to generate documentation. Options:

SVG Diagram - Visual fishbone diagram

HTML Report - Complete analysis with diagram, findings, and recommendations

Both - Full documentation package

JSON Export - Structured data for integration with other tools

Scripts:

```
scripts/generate_diagram.py
```
- Creates SVG fishbone visualization
```
scripts/generate_report.py
```
- Creates HTML report with embedded diagram
```
scripts/export_data.py
```
- Exports analysis data as JSON

目标：生成可视化图表和全面报告。

询问：

准备好生成文档了。选项：

SVG图表 - 可视化鱼骨图

HTML报告 - 包含图表、发现和建议的完整分析报告

两者都要 - 完整文档包

JSON导出 - 结构化数据，用于与其他工具集成

脚本：

```
scripts/generate_diagram.py
```
- 创建SVG鱼骨图可视化
```
scripts/generate_report.py
```
- 创建包含嵌入图表的HTML报告
```
scripts/export_data.py
```
- 将分析数据导出为JSON

Common Pitfalls

常见陷阱

See

references/common-pitfalls.md

for detailed pitfall descriptions and redirection strategies.

Quick reference:

Vague problem statement → Ask for specifics (what/when/where/how much)
Stopping at symptoms → Probe with "Why might this happen?"
Empty categories → Use category-specific prompting questions
Person-blame → Redirect to "What process/system allowed this?"
Groupthink → Use brainwriting before group discussion
Confirmation bias → Challenge assumptions, seek contrary evidence
Too shallow → Add sub-cause levels
Too complex → Consider splitting into multiple diagrams

有关详细的陷阱描述和纠正策略，请参阅

references/common-pitfalls.md

。

快速参考：

模糊的问题陈述 → 询问细节（什么/何时/何地/程度）
停留在症状层面 → 用“为什么会发生这种情况？”深挖
空类别 → 使用类别特定的提示问题
指责个人 → 引导至“什么流程/系统导致了这种情况？”
群体思维 → 在小组讨论前使用书面头脑风暴
确认偏差 → 挑战假设，寻找相反证据
过于浅显 → 添加子原因层级
过于复杂 → 考虑拆分为多个图表

Quality Assessment

质量评估

Rate the analysis on these dimensions (see

references/quality-rubric.md

Dimension	Weight	Description
Problem Clarity	15%	Specific, measurable, non-blaming
Category Coverage	20%	All relevant categories explored
Cause Depth	25%	2-3 levels of sub-causes
Cause Quality	20%	Distinct, actionable, evidence-based
Prioritization	10%	Clear method, justified rankings
Documentation	10%	Complete, visual, shareable

Scoring: Use

scripts/score_analysis.py

to calculate quality score.

从这些维度对分析进行评分（参阅

references/quality-rubric.md

）：

维度	权重	描述
问题清晰度	15%	具体、可衡量、无指责
类别覆盖度	20%	探索所有相关类别
原因深度	25%	2-3层子原因
原因质量	20%	独特、可操作、基于证据
优先级排序	10%	方法清晰，排名合理
文档记录	10%	完整、可视化、可共享

评分：使用

scripts/score_analysis.py

计算质量分数。

Examples

示例

See

references/examples.md

for worked examples:

Manufacturing defect analysis (6Ms)
Customer service complaint (8Ps)
Healthcare incident (4Ss)
Software deployment failure (Custom)

有关已完成的示例，请参阅

references/examples.md

：

制造缺陷分析（6Ms）
客户服务投诉（8Ps）
医疗事件（4Ss）
软件部署失败（自定义）