Graph Schema Validation

Validate knowledge graph schemas and data against defined constraints.

This skill ensures graph data conforms to schema definitions, ontology rules, and consistency requirements.

Quick Start

Use When

• Validating graph data before ingestion
• Testing schema correctness
• Enforcing ontology constraints
• Checking graph consistency
• Verifying RDF/OWL compliance
• Validating property graph models

Inputs

• RDF datasets (Turtle, RDF/XML)
• OWL ontologies
• Property graph schemas
• Cypher structures
• Graph data files
• Graph exports

Outputs

• Validation report
• Detected violations
• Suggested corrections
• Conformance status
• Constraint violations

Example

Input Schema:

Student: student_id (UNIQUE), name, email
Course: course_code (UNIQUE), title, credits
Relationship: (Student)-[:ENROLLED_IN]->(Course)

Input Data:

(Student {name: "Alice"})              # INVALID: missing student_id
(Student {student_id: "S001", name: "Bob"})-[:ENROLLED_IN]->(Course)  # OK

Validation Report:

Violations: 1
- Node S002: Missing required property student_id
- Suggestion: Add student_id to Student node

Validation Types

1. Schema Conformance

• Node labels match schema classes
• Relationships follow schema rules
• Property types are valid

2. Property Validation

• Required properties present
• Data types correct
• Property names consistent

3. Relationship Validation

• Relationships follow schema rules
• Direction correct
• Source/target types valid

4. Cardinality Constraints

• Minimum/maximum occurrences
• Uniqueness constraints
• Collection sizes

5. Graph Integrity

• No orphan nodes
• No broken references
• No duplicate identifiers
• Consistent relationships

Schema Formats

RDF/OWL (SHACL Shapes)

:StudentShape a sh:NodeShape ;
  sh:targetClass :Student ;
  sh:property [
    sh:path :student_id ;
    sh:datatype xsd:string ;
    sh:minCount 1
  ] .

Property Graph (Cypher Constraints)

CREATE CONSTRAINT student_id_unique ON (s:Student) REQUIRE s.student_id IS UNIQUE
CREATE INDEX ON (s:Student)(name)

Execution Steps

1. Load Schema – Load ontology, SHACL shapes, or schema definition
2. Load Data – Load graph data to validate
3. Define Rules – Specify validation constraints
4. Execute Validation – Check data against rules
5. Generate Report – Produce validation results
6. Suggest Fixes – Recommend corrections

Recommended Libraries

• RDF/OWL: rdflib, pyshacl, owlready2
• Property Graph: neo4j, py2neo, networkx
• Data Validation: pydantic, jsonschema
• Graph Analysis: networkx

Best Practices

✓ Validate before production deployment
✓ Enforce constraints at database level
✓ Use consistent property naming
✓ Define validation rules early
✓ Maintain validation tests
✓ Document constraint rules
✓ Review violation reports carefully

References

See validation-patterns.md for validation strategies and example-validations.md for domain validation examples.

Version: 1.0.0