RDF

RDF breaks all information down into atomic subject-predicate-object statements known as triples. This uniform structure allows heterogeneous data from different schemas to be merged without structural conflicts.

• Subject: The resource being described (an IRI or blank node).
• Predicate: The property or relationship (an IRI).
• Object: The value or another resource (an IRI, blank node, or literal).

Example: <http://example.org/contractA> <http://purl.org/dc/terms/date> "2024-01-15"^^xsd:date .

A collection of RDF triples forms a directed, labeled graph. Nodes represent resources or literals, and edges represent named predicates. This graph structure is ideal for representing complex legal networks like citation graphs, corporate ownership structures, and contractual obligation chains.

• Nodes: Subjects and objects.
• Edges: Predicates connecting nodes.
• Labels: IRIs and literals provide semantic meaning.

Unlike relational tables, graph traversal does not require expensive JOIN operations, making it efficient for deep link analysis.

RDF uses Internationalized Resource Identifiers (IRIs) to uniquely identify resources globally. This prevents naming collisions when integrating legal data from multiple jurisdictions or internal systems.

• Global Uniqueness: <https://uscode.house.gov/view.xhtml?req=title:17> unambiguously identifies a specific statute.
• Linked Data: IRIs resolve to machine-readable descriptions (content negotiation).
• Decentralization: Anyone can mint IRIs in their own domain without central registry.

This mechanism is the foundation for linking enterprise legal entities to public authority databases.

RDF graphs are inherently schema-late. Data can be asserted without a predefined schema, and multiple graphs can be merged simply by taking the union of their triples. This is critical for legal knowledge graph construction where data arrives from unstructured contract extraction, structured case management systems, and external regulatory feeds.

• No upfront schema design required.
• Conflict resolution handled at the application or ontology level.
• Open-world assumption: Missing data is treated as unknown, not false.

This contrasts with relational databases where schema changes require migrations.

RDF is an abstract data model with multiple concrete serialization syntaxes optimized for different use cases. Legal systems often exchange data using these standard formats.

• Turtle (TTL): Human-readable, compact syntax for development and debugging.
• RDF/XML: Legacy XML-based syntax for compatibility with older XML toolchains.
• JSON-LD: JSON-based syntax for web APIs and modern JavaScript applications.
• N-Triples/N-Quads: Line-based formats for high-volume streaming and batch processing.

All formats are semantically equivalent, allowing transformation without data loss.

RDF provides mechanisms to make statements about statements, a process called reification. This is essential for legal provenance, where the validity of a fact depends on its source, timestamp, or jurisdiction.

• Standard Reification: Creates a resource representing a triple with rdf:Statement, rdf:subject, rdf:predicate, rdf:object.
• *RDF-star (RDF)**: A modern, concise syntax for directly quoting triples as subjects or objects.

Example: << :contractA :signedBy :Alice >> :witnessedBy :Bob .

This allows attaching confidence scores, extraction sources, and temporal validity to every fact in a legal knowledge graph.