✦ Context Ontology Accelerator

An open-source semantic context layer for AWS that combines knowledge graphs, formal ontologies, and rule-based systems with modern AI — enabling agents to retrieve context, validate it against business logic, and determine correct actions.

Architecture

The system follows a Scan → Model → Serve workflow:

• Scan — Connect data sources, discover schemas, enrich metadata, ingest unstructured documents
• Model — Induce and manage ontologies, define metrics, build a unified semantic graph
• Serve — Query via SPARQL federation (VKG), traverse the knowledge graph, serve context to AI agents via MCP

Access is governed by namespace isolation and role-based access control: namespace-scoped roles (owner, maintainer, data-steward, data-analyst) plus platform-level roles (platform-admin, platform-viewer) that apply across all namespaces. See the control-plane docs for the grants and authorization model.

Quick Start

Prerequisites

• Python 3.12, Node.js 22+, Docker
• pnpm (Node package manager — installed via mise or npm install -g pnpm)
• Java 17+ and Gradle (for Smithy codegen)
• uv (Python package manager)

Setup

Use a release, not the tip of main. We recommend starting from a tagged release — github.com/aws/context-ontology-accelerator/releases. main tracks ongoing development.

# Clone at the latest release tag (replace <tag> with a release from the link above)
git clone --branch <tag> https://github.com/aws/context-ontology-accelerator.git
cd context-ontology-accelerator

make setup      # install Python + CDK TypeScript dependencies
make format     # auto-format code
make lint       # check linting
make test       # run unit tests

Full developer guide: external-docs/content/getting-started.md

Repository Structure

semantic-context/
├── models/              # Smithy API models (source of truth for API contracts)
├── smithy-generated/    # Auto-generated from Smithy (OpenAPI, Python interfaces, TS client)
├── infra/               # AWS CDK (TypeScript) — foundation + per-service stacks
├── packages/
│   ├── control-plane/           # Control Plane APIs
│   ├── data-layer/              # Data Layer APIs (query, retrieval, traversal)
│   ├── sources/                 # Unified data source ingestion (database + documents)
│   ├── ontology-engine/         # Ontology induction, reasoning (HermiT/ELK)
│   ├── metric-service/          # Metric authoring and resolution
│   ├── vkg/                     # Virtual Knowledge Graph (Ontop)
│   ├── mcp-server/              # MCP tools for AI agents
│   ├── context-manager/         # Serve layer Context Manager (query orchestration, upstream clients)
│   └── web-app/                 # React + Cloudscape frontend
├── libs/common/         # Shared Python config, logging, exceptions
├── libs/ts-shared/      # Shared TypeScript interfaces and constants
├── scripts/             # CI-agnostic build/test/deploy scripts
└── external-docs/       # Published documentation (getting-started, deployment guides)

Tech Stack