RDF Semantic Layer

RDF Semantic Layer

Path: rdf/

The RDF semantic layer provides the canonical meaning overlay for Σ OVERWATCH / RAL. While SharePoint (or any document store) remains the artifact store (system of record), RDF supplies stable URIs, typed relationships, and machine-queryable provenance on top of the runtime data.

Standards

Standard	Role
RDF / RDFS / OWL	Core graph model and ontology language
SKOS	Controlled vocabulary alignment
PROV-O	Provenance tracking
SHACL	Shape-based validation (data quality constraints)
SPARQL	Query language for retrieval and proof-of-value

Core Ontology (Core 8 Classes)

The ontology is defined in rdf/ontology/coherence_ops_core.ttl under the namespace ds: <https://deepsigma.ai/ns/coherence#>.

Class	Description
ds:Claim	An assertion made within the system
ds:Evidence	Data or observation that supports or contradicts a claim
ds:SourceArtifact	The upstream document or artifact an evidence item references
ds:DecisionEpisode	A sealed decision event produced by the RAL runtime
ds:Assumption	A premise underlying a decision (with TTL via ds:expiresOn)
ds:Outcome	The result of a decision episode
ds:DriftEvent	A detected deviation from expected behaviour
ds:Patch	A corrective action applied to resolve drift

Key Predicates

Predicate	Domain → Range	Meaning
ds:supports	Evidence → Claim	Evidence backs a claim
ds:contradicts	Evidence → Claim	Evidence undermines a claim
ds:hasEvidence	Claim → Evidence	Claim links to its evidence
ds:basedOn	DecisionEpisode → Assumption	Decision rests on assumption
ds:decides	DecisionEpisode → Claim	Decision resolves a claim
ds:usesSource	Evidence → SourceArtifact	Evidence cites a source
ds:resultedIn	DecisionEpisode → Outcome	Decision produced an outcome
ds:driftedInto	DecisionEpisode → DriftEvent	Decision triggered drift
ds:patchedBy	DriftEvent → Patch	Drift resolved by patch
ds:expiresOn	Assumption → xsd:date	Assumption half-life hook

Directory Layout

Subfolder	Contents
ontology/	Core and extended Turtle ontologies (coherence_ops_core.ttl, coherence_ops_extended.ttl)
shapes/	SHACL constraint graphs (coherence_ops_core.shacl.ttl, coherence_ops_extended.shacl.ttl)
queries/	SPARQL executive query pack (.rq files) plus legacy .sparql queries
mappings/	SharePoint → RDF mapping notes and sample I/O
pipelines/	Extractor and connector contract design docs
examples/	Walkthrough narratives and demo instance graphs
diagrams/	Mermaid architecture diagrams for the RDF layer
sample_data/	Toy Turtle data for local validation and testing

SPARQL Query Pack

The queries in rdf/queries/ demonstrate decision lineage and drift governance:

Query	Purpose
01_expired_assumptions.rq	Find assumptions past their TTL
02_claim_contradictions.rq	Surface claims with contradicting evidence
03_why_retrieval_for_claim.rq	Executive "why" retrieval (target ≤ 60 s)
04_drift_to_patch_trace.rq	Trace drift events to their patches
05_latest_sealed_graphs.rq	List the most recent sealed named graphs

Design Rules

Named Graphs are used for sealing: each sealed decision episode lives in its own immutable named graph. Corrections are applied as patches rather than overwrites, preserving full provenance history.

Canonical vs. Extended

The bundle ships two tiers. The core files (coherence_ops_core.ttl / .shacl.ttl) define the minimum viable ontology and constraints. The extended files add supplementary classes, predicates, and shapes that were merged from the coherence_ops_rdf_module_v0.1 add-on bundle (SharePoint mapping, pipeline contracts, legacy SPARQL, and toy data).

rdf/MANIFEST.md lists every file and its provenance.

See Also

• Coherence Ops Framework — the governance artifacts that RDF triples describe
• Schemas — JSON Schema definitions (complementary to SHACL)
• Sealing and Episodes — how episodes are sealed (named-graph pattern)
• Drift to Patch — drift → patch lifecycle the ontology models
• Concepts — DTE, AL6, contracts, and episodes
• LLM Data Model — data requirements for RAL (graph-first context)
• Mermaid Diagrams — visual documentation

---

Σ OVERWATCH / RAL — Reality Await Layer • Ideas spark. Persistence builds. Deep Sigma delivers.