Knowledge Graph Builder

Knowledge Graph Builder

This skill provides guidance for designing knowledge graphs that capture entities, relationships, and semantic meaning for powerful querying and reasoning.

Core Competencies

• Graph Modeling: Entity-relationship design for graphs
• Query Languages: Cypher (Neo4j), SPARQL (RDF), Gremlin
• Ontology Design: Schema, taxonomies, semantic relationships
• Graph Algorithms: Pathfinding, centrality, community detection

Knowledge Graph Fundamentals

What Makes a Knowledge Graph

Knowledge Graph = Entities + Relationships + Schema + Semantics

Traditional Database:           Knowledge Graph:
┌────────────────────┐         ┌─────────────────────────────┐
│ Tables with rows   │         │ (Person)──KNOWS──▶(Person)  │
│ Foreign keys       │   vs    │     │                       │
│ JOIN operations    │         │   WORKS_AT                  │
│                    │         │     ▼                       │
└────────────────────┘         │ (Company)──IN──▶(Industry)  │
                               └─────────────────────────────┘

When to Use Knowledge Graphs

Graph Data Modeling

Entity Design

Identify core entities (nodes):

// Person entity with properties
CREATE (p:Person {
    id: 'p001',
    name: 'Alice Chen',
    email: 'alice@example.com',
    created_at: datetime()
})

// Multiple labels for categorization
CREATE (c:Organization:Company:TechCompany {
    id: 'c001',
    name: 'Acme Corp',
    founded: 2010
})

Relationship Design

Model connections with typed, directed edges:

// Simple relationship
(person)-[:WORKS_AT]->(company)

// Relationship with properties
(person)-[:WORKS_AT {
    role: 'Engineer',
    start_date: date('2020-01-15'),
    department: 'Engineering'
}]->(company)

// Temporal relationships
(person)-[:EMPLOYED_BY {
    from: date('2018-01-01'),
    to: date('2020-12-31')
}]->(company1)
(person)-[:EMPLOYED_BY {
    from: date('2021-01-01')
}]->(company2)

Common Relationship Patterns

Hierarchical:     (Child)──IS_CHILD_OF──▶(Parent)
                  (Employee)──REPORTS_TO──▶(Manager)

Associative:      (Person)──KNOWS──▶(Person)
                  (Document)──REFERENCES──▶(Document)

Temporal:         (Event)──PRECEDES──▶(Event)
                  (Version)──SUPERSEDES──▶(Version)

Categorical:      (Product)──BELONGS_TO──▶(Category)
                  (Concept)──IS_A──▶(Category)

Spatial:          (Location)──NEAR──▶(Location)
                  (Region)──CONTAINS──▶(City)

Schema Definition

// Node constraints
CREATE CONSTRAINT person_id IF NOT EXISTS
FOR (p:Person) REQUIRE p.id IS UNIQUE;

CREATE CONSTRAINT company_id IF NOT EXISTS
FOR (c:Company) REQUIRE c.id IS UNIQUE;

// Property existence
CREATE CONSTRAINT person_name IF NOT EXISTS
FOR (p:Person) REQUIRE p.name IS NOT NULL;

// Indexes for query performance
CREATE INDEX person_name_idx IF NOT EXISTS
FOR (p:Person) ON (p.name);

CREATE INDEX company_industry_idx IF NOT EXISTS
FOR (c:Company) ON (c.industry);

Cypher Query Patterns

Basic Traversal

// Find all colleagues (people who work at same company)
MATCH (person:Person {name: 'Alice Chen'})-[:WORKS_AT]->(company)
      <-[:WORKS_AT]-(colleague:Person)
WHERE colleague <> person
RETURN colleague.name, company.name

// Variable-length paths (1-3 hops)
MATCH path = (start:Person)-[:KNOWS*1..3]->(end:Person)
WHERE start.name = 'Alice Chen' AND end.name = 'Bob Smith'
RETURN path, length(path) as hops

Aggregation

// Count relationships
MATCH (p:Person)-[:WORKS_AT]->(c:Company)
RETURN c.name, count(p) as employee_count
ORDER BY employee_count DESC

// Collect into lists
MATCH (p:Person)-[:HAS_SKILL]->(s:Skill)
RETURN p.name, collect(s.name) as skills

Recommendations

// "People you may know" - friends of friends
MATCH (me:Person {id: $userId})-[:KNOWS]-(friend)-[:KNOWS]-(suggestion)
WHERE NOT (me)-[:KNOWS]-(suggestion) AND me <> suggestion
RETURN suggestion.name, count(friend) as mutual_friends
ORDER BY mutual_friends DESC
LIMIT 10

// Content-based: similar interests
MATCH (me:Person {id: $userId})-[:INTERESTED_IN]->(topic)
      <-[:INTERESTED_IN]-(similar:Person)
WHERE me <> similar
WITH similar, count(topic) as shared_interests
ORDER BY shared_interests DESC
RETURN similar.name, shared_interests
LIMIT 10

Path Analysis

// Shortest path
MATCH path = shortestPath(
    (start:Person {name: 'Alice'})-[:KNOWS*]-(end:Person {name: 'Bob'})
)
RETURN path, length(path)

// All shortest paths
MATCH path = allShortestPaths(
    (start:Person)-[:KNOWS*]-(end:Person)
)
WHERE start.name = 'Alice' AND end.name = 'Bob'
RETURN path

Graph Algorithms

Centrality Measures

// Using Neo4j Graph Data Science
CALL gds.pageRank.stream('myGraph')
YIELD nodeId, score
RETURN gds.util.asNode(nodeId).name AS name, score
ORDER BY score DESC
LIMIT 10

Community Detection

// Louvain for community detection
CALL gds.louvain.stream('myGraph')
YIELD nodeId, communityId
RETURN communityId, collect(gds.util.asNode(nodeId).name) as members
ORDER BY size(members) DESC

Knowledge Graph Patterns

Entity Resolution

// Find potential duplicates
MATCH (p1:Person), (p2:Person)
WHERE p1.id < p2.id
  AND (p1.email = p2.email
       OR (p1.name = p2.name AND p1.birth_date = p2.birth_date))
RETURN p1, p2

// Merge duplicates
MATCH (p1:Person {id: 'keep'}), (p2:Person {id: 'duplicate'})
CALL apoc.refactor.mergeNodes([p1, p2], {
    properties: 'combine',
    mergeRels: true
})
YIELD node
RETURN node

Semantic Layering

┌─────────────────────────────────────────────────────┐
│                 Instance Layer                       │
│   (Alice)──KNOWS──▶(Bob)                            │
│   (Alice)──WORKS_AT──▶(Acme)                        │
├─────────────────────────────────────────────────────┤
│                  Schema Layer                        │
│   (:Person)──CAN_KNOW──▶(:Person)                   │
│   (:Person)──CAN_WORK_AT──▶(:Company)               │
├─────────────────────────────────────────────────────┤
│                 Ontology Layer                       │
│   (Person)──IS_A──▶(Agent)                          │
│   (Company)──IS_A──▶(Organization)                  │
└─────────────────────────────────────────────────────┘

Temporal Modeling

// State over time
CREATE (person)-[:HAS_STATE {
    valid_from: date('2020-01-01'),
    valid_to: date('2020-12-31')
}]->(state:PersonState {
    status: 'employed',
    salary: 80000
})

// Query state at point in time
MATCH (p:Person {id: $personId})-[r:HAS_STATE]->(s)
WHERE r.valid_from <= date($queryDate)
  AND (r.valid_to IS NULL OR r.valid_to >= date($queryDate))
RETURN s

Best Practices

Modeling Guidelines

1. Prefer relationships over properties when the connection has meaning
2. Use specific relationship types (:MANAGES not :RELATED_TO)
3. Model for your queries - understand access patterns first
4. Keep properties atomic - no arrays for searchable data
5. Version nodes, not graphs - temporal properties on relationships

Performance Tips

• Index properties used in WHERE clauses
• Use parameters ($userId) not string concatenation
• Limit variable-length paths (*1..5 not *)
• Profile queries with EXPLAIN and PROFILE
• Consider relationship direction in traversals

References

• references/cypher-patterns.md - Advanced Cypher query examples
• references/graph-modeling.md - Entity and relationship design patterns
• references/graph-algorithms.md - Algorithm selection and configuration