Abstract blue and silver data ribbons flowing through dark space

Release version 1.0

Ontology-native · distributed · exact by construction

c8

Next Generation RDF DB

A database for organizations whose data must carry meaning—not just values. c8 combines governed RDF, a Release 1.0 target of full applicable SPARQL 1.1 conformance, OWL 2 DL reasoning and distributed execution behind one evidence-bound query plane.

Understand the value Configure & deploy

Business

Why semantic infrastructure changes what the organization can safely ask and automate.

Operators

How to load governed TriG and deploy highly available RKE2/K3s, EKS, AKS, GKE or on-prem clusters.

Engineers

How Rust, Java/HermiT, future C++ kernels, Arrow, Parquet and Kubernetes fit together.

The business case

Turn shared meaning into usable infrastructure.

Traditional data platforms can store relationships. c8 is being built to preserve what those relationships mean, prove when an answer is complete, and distribute the work without changing the result.

Ask questions across meaning, not tables

Model classes, relationships, constraints and cross-domain semantics once, then query the governed graph through SPARQL instead of rebuilding business meaning in every application.

Move expensive reasoning out of the hot path

Use immutable semantic indexes and qualified finite materializations for selective workloads, while retaining an exact HermiT path whenever optimized coverage cannot prove completeness.

Make every answer defensible

Bind query results to authorization, active snapshot, graph set, ontology versions, checksums, proofs and completeness evidence. Unknown coverage is never treated as false.

What the organization gets

Fewer semantic silos

Connect customer, product, operational and evidence domains without flattening their meaning.

Safer automation

Give applications and agents answers constrained by governed ontology and graph authorization.

Predictable scale

Partition sparse graph work across bounded workers and add role-specific Kubernetes capacity as demand grows.

Open standards

Build on RDF 1.1 TriG, SPARQL 1.1, OWL 2 DL, REST, OpenAPI, Arrow and Parquet.

The eight values of a reasoning system

The 8 Cs of c8.

The name stands for eight properties a business-grade reasoning system should make operational. Together they turn governed facts into certified context before that context reaches predictive AI.

Deterministic meaning first. Predictive intelligence second.

c8 reasons over explicit OWL axioms. It does not predict the answer; it derives what follows from the authorized ontology and returns the context plus evidence to applications or models.

Context

Build the smallest authorized semantic graph that answers the question, instead of sending an AI model an undifferentiated data dump.

Correctness

Evaluate RDF terms, SPARQL algebra and OWL entailment under explicit standards semantics—not an approximate similarity score.

Consistency

Reject publication when the checksum-bound ontology snapshot is globally inconsistent under HermiT.

Completeness

Return success only after every required partition and reasoning route proves that the answer set is whole.

Constraints

Carry class restrictions, disjointness, keys, cardinality and property characteristics into the meaning of the data.

Connectivity

Traverse governed relationships across named subdomains while preserving graph identity and authorization.

Computation

Distribute joins, property paths and finite semantic materialization across bounded workers without changing the scalar result.

Certification

Bind answers to snapshot, graph-set, plan, checksum, proof-support and completeness evidence.

What it is for

A governed meaning layer for consequential decisions.

Use c8 after data has been mapped into ontology-grounded TriG. It becomes the versioned semantic system that applications, analysts and agents can query without each reinterpreting the enterprise model.

Trusted AI context

Ground agents and automation in authorized entity relationships, ontology rules and evidence-bound answers.

Cross-domain operations

Traverse customer, product, logistics, maintenance and risk domains through stable IRIs and governed links.

Explainable compliance

Connect an answer to its active snapshot, graph authorization, ontology versions, support and completeness status.

Semantic application platform

Give RDF clients, BI tools and services a standard SPARQL interface plus an enterprise JSON evidence API.

One correctness boundary

Fast where certified. Exact where required.

The planner uses the fastest execution lane that can prove completeness for the exact request. Incomplete or unknown semantic coverage routes to HermiT; it is never silently interpreted as “no answer.”

SPARQL + REST

query, protocol, Swagger

Semantic planner

graphs, algebra, coverage

Exact reasoning

OWL Direct + HermiT

Distributed runtime

Arrow, Grace, spill

RDF + Parquet

immutable snapshots

Certified context graph generation

Prepared TriG

A/T/R/C axioms · named graphs

Qualified snapshot

OWL 2 DL · pinned imports

SPARQL + reasoning

distributed algebra · exact fallback

Context graph

authorized facts · inferences · proof

A context graph is the minimum authorized semantic slice needed for a request: asserted triples, entailed relationships, relevant constraints, named-graph identity and proof/support references. Optional hydration can attach enterprise payload columns after the semantic entities are selected.

Execution model

Relevant graph routing → typed algebra → authorized semantic slice → stable partitions → vectorized joins and property paths → exact completeness/proof validation → GUID-directed Parquet hydration.

Fail closed by default

Missing partitions, bad checksums, stale snapshots, unauthorized graphs and incomplete proofs cannot become successful answers.

Implementation advantage

Use each language at the right trust boundary.

The code review found 95 Rust source files, three Java sources in the HermiT adapter and no first-party C or C++ files. Release messaging reflects that evidence.

Rust · active

Memory-safe services, Kubernetes controllers, bounded concurrency, Arrow/Parquet exchange, Grace joins, caching, spill and deterministic graph work units.

Java + HermiT · active

A checksum-pinned OWLAPI/HermiT adapter validates OWL 2 DL, checks consistency and executes exact Direct-Semantics entailment where certified indexes cannot prove coverage.

C++ · planned acceleration lane

A future, benchmark-gated boundary for measured CSR, bitmap and SIMD sparse kernels. The current candidate contains no first-party C++ implementation and keeps OpenMP/BLAS at one thread for sparse RDF work.

Release 1.0 standards contract

The target is full applicable SPARQL 1.1 query, protocol and result-format conformance, with SPARQL evaluated under the OWL 2 Direct Semantics entailment regime over a validated OWL 2 DL snapshot. That claim remains qualification-gated by W3C suites and Apache Jena differential tests.

Accuracy over slogans

“OWL-DL-compliant SPARQL” is useful shorthand, but the precise statement separates the ontology profile from the query entailment regime. The current scalar path protects correctness while distributed operators finish activation and qualification.

What “100% SPARQL 1.1” means for Release 1.0

Every query form

SELECT, ASK, CONSTRUCT and DESCRIBE with standards-correct RDF term and bag semantics.

Complete algebra

JOIN, OPTIONAL, UNION, MINUS, FILTER, BIND, VALUES, subqueries, aggregation, ordering, slicing and property paths.

Protocol surface

Dataset parameters, result negotiation, service description and secured SERVICE/SERVICE SILENT federation.

Executable evidence

Applicable W3C suites plus Apache Jena differential results; optimized output must equal the scalar oracle.

Current truth: the implementation already covers substantial SPARQL parsing, scalar evaluation, typed distributed plans and native operator kernels, but federation, live worker activation and full release-suite evidence remain open. “100%” is the Release 1.0 acceptance target, not a passed claim for the present candidate.

Deploy c8

Follow the RKE2/K3s, EKS, AKS, GKE or on-prem Kubernetes sequence; size responsibility pools; load governed TriG; publish a certified snapshot.

Open Helm guide

Contribute

Navigate the architecture by feature and trace every subsystem to first-party Rust, Java, contracts, charts and qualification evidence.

Explore the source

Know the boundary

The implementation is substantial, but native builds, federation, worker activation and real multinode qualification remain release gates.

Read qualification status