Keystone Overview

Keystone is a knowledge graph platform for RDF storage, SPARQL querying, semantic search, and structured knowledge management — available as a Naburis product.

Architecture

Keystone is composed of specialized microservices, each responsible for a distinct capability:

Service	Description
Gateway	API entry point, auth, routing, SPARQL proxy
Statement	Triple store operations (CRUD on RDF statements)
Projection	Materialized views and graph projections
Index	Full-text and vector search indexing
Catalog	Dataset cataloging (DCAT) and metadata management
Workflow	ETL pipeline orchestration
Ingestion	Bulk data import (CSV, JSON-LD, RDF, etc.)
Ontology	OWL ontology management and reasoning
Shapes	SHACL shape validation
Semantic	Embedding generation and similarity search
Oracle	AI-powered knowledge assistance
NLQuery	Natural language to SPARQL translation
Translate	Multilingual content translation

Data Model

Keystone stores data as RDF triples (subject → predicate → object) organized into named graphs. Each tenant’s data is isolated by context ID (named graph URI).

Multi-Tenancy

Tenant isolation is enforced at three levels:

1. JWT Claims — Each request carries X-Tenant-ID validated against the JWT
2. Named Graphs — SPARQL queries are scoped to the tenant’s named graph
3. Row-Level — Database queries filter by tenant where applicable

Key Endpoints

All authenticated endpoints require Authorization, X-Tenant-ID, X-Workspace-ID, and X-Context-ID headers.

Method	Path	Description
POST	/sparql	SPARQL SELECT / ASK / CONSTRUCT queries
POST	/sparql/update	SPARQL UPDATE / INSERT / DELETE operations
POST	/gql	Graph traversal queries (nodes + edges)
GET	/health	Service health check
GET	/ops/health	Deep health check (all services + dependencies)
GET	/auth/config	OIDC configuration for SPA clients

See the full API Reference for request/response examples.