wiki-disambiguate

name	wiki-disambiguate
description	Disambiguate entities and manage aliases for the knowledge wiki
allowed-tools	["Read","Grep","Bash","Edit","AskUserQuestion"]

Wiki Disambiguate Skill

Manage entity aliases and resolve naming conflicts in the wiki. Uses Obsidian CLI's alias system to handle synonyms, abbreviations, and multi-language terms.

Usage

/wiki-disambiguate                    # Interactive: scan for disambiguation opportunities
/wiki-disambiguate --auto             # Auto-add common aliases
/wiki-disambiguate "RAG"              # Disambiguate specific term
/wiki-disambiguate --merge            # Find and merge duplicate entities

What is Disambiguation?

Synonyms (Same entity, different names)

[[RAG]] → [[Retrieval-Augmented Generation]]
[[LLM]] → [[Large Language Models]]
[[大语言模型]] → [[Large Language Models]]

Polysemy (Same name, different entities)

[[Apple]] → Company? Fruit?
[[Transformer]] → Architecture? Movie? Device?

Solution: Aliases

---
aliases: ["RAG", "检索增强生成"]
---
# Retrieval-Augmented Generation

With aliases, [[RAG]] automatically resolves to [[Retrieval-Augmented Generation]].

Workflow

Step 1: Scan for potential duplicates

Find entities that might be the same concept:

# Get all entities
obsidian files folder=wiki/entities format=json

Detection strategies:

• Similar names (edit distance, singular/plural)
• One name is abbreviation of another
• Same type + shared sources

Step 2: Compare entity pairs

# Read both entities
obsidian read path="wiki/entities/EntityA.md"
obsidian read path="wiki/entities/EntityB.md"

Compare:

• Content overlap
• Same type
• Shared [[source]] references

Step 3: Interactive disambiguation

Potential duplicate detected:

[[Entity A]] (primary)
- Type: abstract
- Facts: 3

[[Entity B]] (similar)
- Type: abstract
- Facts: 2

Actions:
A) Link B to A (add alias + reference in A)
B) Link A to B (add alias + reference in B)
C) Keep separate (different concepts)
D) Skip

Step 4: Execute linking

When user chooses to link:

# 1. Add alias to primary entity
obsidian property:set name="aliases" value="EntityB" file="EntityA"

# 2. Add reference line in primary entity
# Edit EntityA to add:
# > See also: [[EntityB]]

Result:

• [[EntityB]] resolves to EntityA via alias
• EntityB file kept intact (no broken links)
• Cross-references preserved

Step 5: Update cache

ls wiki/entities/*.md | xargs -I {} basename {} .md > wiki/cache.md

Common Alias Patterns

Academic/Technical Terms

Entity Name	Suggested Aliases
Retrieval-Augmented Generation	RAG, 检索增强生成
Large Language Models	LLM, 大语言模型
Chain of Thought	CoT, 思维链
Supervised Fine-Tuning	SFT, 监督微调
Reinforcement Learning	RL, 强化学习
Knowledge Graphs	KG, 知识图谱
Markov Decision Process	MDP, 马尔可夫决策过程

Person Names

Entity Name	Suggested Aliases
Geoffrey Hinton	Hinton, 杰弗里·辛顿
Andrew Ng	Ng, 吴恩达

Organization/Project Names

Entity Name	Suggested Aliases
OpenAI	openai
DeepSeek	deepseek, 深度求索

Disambiguation Report

## Disambiguation Report

### Aliases Added
- [[Retrieval-Augmented Generation]] ← `RAG`, `检索增强生成`
- [[Large Language Models]] ← `LLM`, `大语言模型`
- [[Chain of Thought]] ← `CoT`

### Potential Duplicates (needs review)
- [[Knowledge Graph]] vs [[Knowledge Graphs]] — similar content, consider merging
- [[Language Model]] vs [[Large Language Models]] — different scope, keep separate

### Polysemy Detected
- [[Transformer]] — multiple meanings detected:
  - [[Transformer (Architecture)]] — neural network architecture
  - Suggest creating separate pages with disambiguation page

Obsidian CLI Commands Reference

Command	Purpose
obsidian aliases	List all aliases in vault
obsidian aliases file="X"	Get aliases for specific entity
obsidian property:set name="aliases" value='[...]' file="X"	Set aliases
obsidian property:read name="aliases" file="X"	Read aliases
obsidian search query="X"	Find potential synonym matches

Workflow Examples

Example 1: Auto-add common abbreviations

# Detect LLM-related entities
obsidian search query="Large Language Model" format=json

# Add alias
obsidian property:set name="aliases" value="LLM" file="Large Language Models"

Example 2: Merge duplicate entities

Detected potential duplicates:
- [[Knowledge Graph]] (47 words)
- [[Knowledge Graphs]] (32 words)

Both describe structured data representation.

Merge action:
1. Keep [[Knowledge Graphs]] as primary (better fit with plural convention)
2. Add alias "Knowledge Graph"
3. Merge content
4. Delete [[Knowledge Graph]]

Example 3: Handle polysemy

Found multiple meanings for "Transformer":

1. [[Transformer (Architecture)]] — neural network, attention mechanism
2. No page for: Transformer (Electrical) — power distribution device

Actions:
A) Rename to [[Transformer (Architecture)]], add alias "Transformer"
B) Keep as is, note that it's AI-focused
C) Create disambiguation page listing all meanings

Edge Cases

Circular Aliases

Avoid: Entity A has alias B, Entity B has alias A.

Wrong:
  [[RAG]] aliases: ["Retrieval-Augmented Generation"]
  [[Retrieval-Augmented Generation]] aliases: ["RAG"]

Correct:
  [[Retrieval-Augmented Generation]] aliases: ["RAG"]
  No separate [[RAG]] page needed

Alias Conflicts

If alias already exists on different entity:

Conflict: "CoT" already aliased to [[Chain of Thought]]

New candidate: [[Chain-of-Thought Reasoning]] also wants "CoT"

Options:
1. Keep current assignment
2. Move alias to new entity
3. Use longer form "CoTR" for new entity

Partial Name Matching

Query: "language model"
Results:
- [[Large Language Models]] ← add alias "Language Model"?
- [[Language Model]] ← separate entity?

Decision: Different scopes (large vs general), keep separate