Skip to content

A Day in the Life: Knowledge Graph Personas

Personas: OA (Ontology Architect), KB (Knowledge Base Engineer), SDE (Semantic Data Engineer)

Morning: Ontology Design

OA reviews a request to model a new domain in the FCC knowledge graph. The domain -- regulatory compliance -- needs classes for regulations, obligations, evidence artifacts, and compliance statuses. OA designs the ontology schema: class hierarchies (Regulation > Obligation > Control), object properties (hasObligation, evidencedBy, appliesTo), and data properties (effectiveDate, jurisdiction, severityLevel).

OA validates the schema against existing ontology patterns to ensure consistency. The FCC knowledge graph already defines 9 node types and 9 edge types; the new domain classes must integrate cleanly with the existing vocabulary. OA produces an ontology specification document with class diagrams and property definitions.

Midday: Graph Population

KB takes OA's schema and populates the knowledge graph. KB ingests data from multiple sources: persona YAML files, workflow graph JSON definitions, governance rule YAML, and cross-reference matrices. Each data element becomes a typed node with properties, and each relationship becomes a typed edge.

KB uses the build_full_fcc_graph() builder to generate the base graph, then enriches it with the new regulatory compliance nodes. KB runs validation checks: are all required properties present? Do all edge endpoints reference existing nodes? Are there any orphaned nodes with no connections?

Afternoon: Serialization and Querying

SDE implements the data access layer. SDE configures serializers for multiple output formats: OWL/RDF for integration with triple stores like Apache Jena, SKOS for vocabulary management systems, JSON-LD for web applications, and custom JSON for the Streamlit visualization apps.

SDE also builds query interfaces: given a persona ID, return all connected artifacts within two hops; given a regulation, return all personas whose constitutions reference it; given a workflow node, return the upstream and downstream dependency chain. SDE optimizes query performance by building indexes on frequently queried properties.

Tools Used

  • KnowledgeGraph for graph operations
  • build_full_fcc_graph() and other builder functions
  • OWLSerializer, RDFSerializer, SKOSSerializer, JSONLDSerializer
  • FCCOntology for schema modeling
  • FederatedKnowledgeGraph for cross-namespace graphs

Key Outputs

  • Ontology schemas with class hierarchies and properties (OA)
  • Populated knowledge graphs with validation reports (KB)
  • Multi-format serialized graph exports (SDE)
  • Query interfaces and access layer implementations (SDE)