HR Network Analyst
Applies graph theory and network science to professional relationship mapping. Identifies hidden superconnectors, influence brokers, and knowledge mavens that drive professional ecosystems.
Integrations
Works with: career-biographer, competitive-cartographer, research-analyst, cv-creator
Core Questions Answered
- Who should I know? (optimal networking targets)
- Who knows everyone? (superconnectors for referrals)
- Who bridges worlds? (cross-domain brokers)
- How does influence flow? (information/opportunity pathways)
- Where are structural holes? (untapped connection opportunities)
Quick Start
User: "Who are the key connectors in AI safety research?"
Process:
1. Define boundary: AI safety researchers, 2020-2024
2. Identify sources: arXiv, NeurIPS workshops, Twitter clusters
3. Compute centrality: betweenness (bridges), eigenvector (influence)
4. Classify by archetype: Connector, Maven, Broker
5. Output: Ranked list with network position rationale
Key principle: Most valuable people aren't always most famous—they connect otherwise disconnected worlds.
Gladwellian Archetypes (Quick Reference)
| Type |
Network Signature |
HR Value |
| Connector |
High betweenness + degree, bridges clusters |
Best for cross-domain referrals |
| Maven |
High in-degree, authoritative, creates content |
Know who's good at what |
| Salesman |
High influence propagation, deal networks |
Close candidates, navigate negotiation |
Full theory: See references/network-theory.md
Centrality Metrics (Quick Reference)
| Metric |
Meaning |
When to Use |
| Betweenness |
Controls information flow |
Finding gatekeepers, brokers |
| Degree |
Raw connection count |
Maximizing referral reach |
| Eigenvector |
Quality over quantity |
Access to power, rising stars |
| PageRank |
Endorsed by important others |
Thought leaders |
| Closeness |
Can reach anyone quickly |
Information spreading |
Analysis Workflows
1. Find Superconnectors for Referrals
- Define target domain → Seed network → Expand → Compute betweenness + degree → Rank
2. Map Domain Influence
- Define boundaries → Multi-source construction → Community detection → Identify brokers
3. Optimize Personal Networking
- Map current network → Map target domain → Find shortest paths → Identify structural holes
4. Organizational Network Analysis (ONA)
- Collect data (surveys, Slack metadata) → Construct graph → Find informal vs formal structure
Detailed workflows: See references/data-sources-implementation.md
Data Sources
| Source |
Signal Strength |
What to Extract |
| Co-authorship |
Very strong |
Publication collaborations |
| Conference co-panel |
Strong |
Speaking relationships |
| GitHub co-repo |
Medium-strong |
Code collaboration |
| LinkedIn connection |
Medium |
Professional links |
| Twitter mutual |
Weak |
Social association |
Multi-source fusion: Weight and combine signals for robust network
When NOT to Use
- Surveillance: Tracking individuals without consent
- Discrimination: Using network position to exclude
- Manipulation: Engineering social influence for harm
- Privacy violation: Accessing non-public data
- Speculation without data: Guessing network structure
Anti-Patterns
Anti-Pattern: Degree Obsession
What it looks like: Only looking at who has most connections
Why wrong: High degree often = noise; connectors differ from popular
Instead: Use betweenness for bridging, eigenvector for influence quality
Anti-Pattern: Static Network Assumption
What it looks like: Treating 5-year-old connections as current
Why wrong: Networks evolve; old edges may be dead
Instead: Recency-weight edges, verify currency
Anti-Pattern: Single-Source Reliance
What it looks like: Using only LinkedIn data
Why wrong: Missing relationships not on LinkedIn
Instead: Multi-source fusion with source-appropriate weighting
Anti-Pattern: Ignoring Context
What it looks like: High betweenness = valuable, regardless of domain
Why wrong: Bridging irrelevant communities isn't useful
Instead: Constrain analysis to relevant domain boundaries
Ethical Guidelines
Acceptable:
- Analyzing public data (conference speakers, publications)
- Aggregate pattern analysis
- Opt-in organizational analysis
- Academic research with proper IRB
NOT Acceptable:
- Scraping private profiles without consent
- Building surveillance systems
- Selling individual data
- Discrimination based on network position
Troubleshooting
| Issue |
Cause |
Fix |
| Can't find data |
Domain small/private |
Snowball sampling, surveys, adjacent communities |
| False edges |
Over-weighting weak signals |
Require multiple signals, threshold weights |
| Too large |
Unconstrained boundary |
K-core filtering, high-weight only |
| Entity resolution |
Same person, different names |
Unique IDs (ORCID), manual verification |
Reference Files
references/algorithms.md - NetworkX code patterns, centrality formulas, Gladwell classification
references/graph-databases.md - Neo4j, Neptune, TigerGraph, ArangoDB query examples
references/data-sources.md - LinkedIn network data acquisition strategies, APIs, scraping, legal considerations
Core insight: Advantage comes from bridging otherwise disconnected groups, not from connections within dense clusters. — Ron Burt, Structural Holes Theory
1---2name: hr-network-analyst-23description: Professional network graph analyst identifying Gladwellian superconnectors, mavens, and influence brokers using betweenness centrality, structural holes theory, and multi-source network reconstruction.4 Activate on 'superconnectors', 'network analysis', 'who knows who', 'professional network', 'influence mapping', 'betweenness centrality'. NOT for surveillance, discrimination, stalking, privacy violation,5 or speculation without data.6allowed-tools: Read,Write,Edit,WebSearch,WebFetch,mcp__firecrawl__firecrawl_search,mcp__firecrawl__firecrawl_scrape,mcp__brave-search__brave_web_search,mcp__SequentialThinking__sequentialthinking7metadata:8 source: https://skillmd.com/skills/curiositech/hr-network-analyst-29 category: Research & Analysis10 pairs-with:11 - skill: career-biographer12 reason: Understand network in career context13 - skill: competitive-cartographer14 reason: Map competitive professional landscape15 tags:16 - network17 - superconnectors18 - influence19 - graph-theory20 - hr21---2223# HR Network Analyst2425Applies graph theory and network science to professional relationship mapping. Identifies hidden superconnectors, influence brokers, and knowledge mavens that drive professional ecosystems.2627## Integrations2829Works with: career-biographer, competitive-cartographer, research-analyst, cv-creator3031## Core Questions Answered3233- **Who should I know?** (optimal networking targets)34- **Who knows everyone?** (superconnectors for referrals)35- **Who bridges worlds?** (cross-domain brokers)36- **How does influence flow?** (information/opportunity pathways)37- **Where are structural holes?** (untapped connection opportunities)3839## Quick Start4041```42User: "Who are the key connectors in AI safety research?"4344Process:451. Define boundary: AI safety researchers, 2020-2024462. Identify sources: arXiv, NeurIPS workshops, Twitter clusters473. Compute centrality: betweenness (bridges), eigenvector (influence)484. Classify by archetype: Connector, Maven, Broker495. Output: Ranked list with network position rationale50```5152**Key principle**: Most valuable people aren't always most famous—they connect otherwise disconnected worlds.5354## Gladwellian Archetypes (Quick Reference)5556| Type | Network Signature | HR Value |57|------|-------------------|----------|58| **Connector** | High betweenness + degree, bridges clusters | Best for cross-domain referrals |59| **Maven** | High in-degree, authoritative, creates content | Know who's good at what |60| **Salesman** | High influence propagation, deal networks | Close candidates, navigate negotiation |6162**Full theory**: See `references/network-theory.md`6364## Centrality Metrics (Quick Reference)6566| Metric | Meaning | When to Use |67|--------|---------|-------------|68| **Betweenness** | Controls information flow | Finding gatekeepers, brokers |69| **Degree** | Raw connection count | Maximizing referral reach |70| **Eigenvector** | Quality over quantity | Access to power, rising stars |71| **PageRank** | Endorsed by important others | Thought leaders |72| **Closeness** | Can reach anyone quickly | Information spreading |7374## Analysis Workflows7576### 1. Find Superconnectors for Referrals77- Define target domain → Seed network → Expand → Compute betweenness + degree → Rank7879### 2. Map Domain Influence80- Define boundaries → Multi-source construction → Community detection → Identify brokers8182### 3. Optimize Personal Networking83- Map current network → Map target domain → Find shortest paths → Identify structural holes8485### 4. Organizational Network Analysis (ONA)86- Collect data (surveys, Slack metadata) → Construct graph → Find informal vs formal structure8788**Detailed workflows**: See `references/data-sources-implementation.md`8990## Data Sources9192| Source | Signal Strength | What to Extract |93|--------|-----------------|-----------------|94| Co-authorship | Very strong | Publication collaborations |95| Conference co-panel | Strong | Speaking relationships |96| GitHub co-repo | Medium-strong | Code collaboration |97| LinkedIn connection | Medium | Professional links |98| Twitter mutual | Weak | Social association |99100**Multi-source fusion**: Weight and combine signals for robust network101102## When NOT to Use103104- **Surveillance**: Tracking individuals without consent105- **Discrimination**: Using network position to exclude106- **Manipulation**: Engineering social influence for harm107- **Privacy violation**: Accessing non-public data108- **Speculation without data**: Guessing network structure109110## Anti-Patterns111112### Anti-Pattern: Degree Obsession113**What it looks like**: Only looking at who has most connections114**Why wrong**: High degree often = noise; connectors differ from popular115**Instead**: Use betweenness for bridging, eigenvector for influence quality116117### Anti-Pattern: Static Network Assumption118**What it looks like**: Treating 5-year-old connections as current119**Why wrong**: Networks evolve; old edges may be dead120**Instead**: Recency-weight edges, verify currency121122### Anti-Pattern: Single-Source Reliance123**What it looks like**: Using only LinkedIn data124**Why wrong**: Missing relationships not on LinkedIn125**Instead**: Multi-source fusion with source-appropriate weighting126127### Anti-Pattern: Ignoring Context128**What it looks like**: High betweenness = valuable, regardless of domain129**Why wrong**: Bridging irrelevant communities isn't useful130**Instead**: Constrain analysis to relevant domain boundaries131132## Ethical Guidelines133134**Acceptable**:135- Analyzing public data (conference speakers, publications)136- Aggregate pattern analysis137- Opt-in organizational analysis138- Academic research with proper IRB139140**NOT Acceptable**:141- Scraping private profiles without consent142- Building surveillance systems143- Selling individual data144- Discrimination based on network position145146## Troubleshooting147148| Issue | Cause | Fix |149|-------|-------|-----|150| Can't find data | Domain small/private | Snowball sampling, surveys, adjacent communities |151| False edges | Over-weighting weak signals | Require multiple signals, threshold weights |152| Too large | Unconstrained boundary | K-core filtering, high-weight only |153| Entity resolution | Same person, different names | Unique IDs (ORCID), manual verification |154155## Reference Files156157- `references/algorithms.md` - NetworkX code patterns, centrality formulas, Gladwell classification158- `references/graph-databases.md` - Neo4j, Neptune, TigerGraph, ArangoDB query examples159- `references/data-sources.md` - LinkedIn network data acquisition strategies, APIs, scraping, legal considerations160161---162163**Core insight**: Advantage comes from bridging otherwise disconnected groups, not from connections within dense clusters. — Ron Burt, Structural Holes Theory