Open Data Product Graphs

A vendor-neutral, machine-readable graph model for connecting data products, use cases, objectives, KPIs, policies, APIs, agents, and business value.

---

What Is ODPG?

Open Data Product Graphs (ODPG) is the relationship and intelligence layer of the Open Data Products ecosystem. It gives organizations a standard way to describe how enterprise assets connect across strategy, governance, semantics, operations, and AI-assisted reasoning.

ODPG helps answer questions such as:

• Which data products support a business objective?
• Which use cases depend on a data product, API, workflow, or agent?
• Which KPIs measure strategic outcomes?
• Which policies govern downstream assets?
• Where are strategic gaps, overlaps, or unsupported objectives?
• What trusted graph context should an AI agent use before acting?

Why It Matters

Modern data ecosystems are distributed across catalogs, platforms, teams, policies, agents, APIs, and business domains. ODPG connects those fragments into a graph-native structure that is readable by humans, software, and AI systems.

Capability	What ODPG Enables
Strategic alignment	Link use cases, data products, KPIs, and objectives
Governance reasoning	Trace policy relationships across connected assets
Impact analysis	Follow dependencies and relationship paths
AI grounding	Provide trusted graph context for agents and automation
Interoperability	Reference ODPS, ODPV, ODPC, and external graph formats
Visualization	Generate graph explorer views for inspection and review

Repository Contents

Path	Purpose
source/index.html.md	Main human-readable ODPG specification
source/includes/	Specification sections rendered into the docs site
source/schema/odpg.yaml	YAML schema for ODPG graph validation
source/schema/odpg.json	JSON schema for ODPG graph validation
source/graph/objects.jsonl	Agent-friendly graph resource for retrieval and reasoning
source/examples/	Example ODPG graphs and related assets
source/scripts/	Python toolkit for validation, conversion, traversal, analysis, and visualization
tests/	Pytest coverage for the shared toolkit logic
build/	Generated static documentation output

Quick Start

Clone the repository and install the documentation dependencies:

git clone https://github.com/Open-Data-Product-Initiative/odpg-v1.0.git
cd odpg-v1.0
bundle install

Run the local documentation site:

bundle exec middleman serve

Open the docs at:

http://localhost:4567

Build the static site:

bundle exec middleman build

Python Toolkit

Install development dependencies:

python -m pip install -r requirements-dev.txt

Validate an ODPG graph:

python source/scripts/odpg_validate.py -i source/examples/SLA/sla.yml

Convert an external graph format into ODPG YAML:

python source/scripts/odpg_convert.py -i path/to/source.graphml -o graph.yml

Summarize graph metadata, nodes, edges, and confidence values:

python source/scripts/odpg_summary.py -i graph.yml

Traverse relationships from a starting node:

python source/scripts/odpg_traverse.py -i graph.yml --start DP-001 --depth 2

Analyze governance and strategic gaps:

python source/scripts/odpg_analyze.py -i graph.yml

Extract trusted context around a node for AI-agent use:

python source/scripts/odpg_agent_context.py -i graph.yml --node AGENT-001 --depth 2

Generate a standalone graph explorer:

python source/scripts/generate_graph_explorer.py -i graph.yml -o graph-explorer.html

Minimal Graph Example

schema: https://opendataproducts.org/odpg-v1.0/schema/odpg.yaml
version: "1.0"
kind: Graph
graph:
  metadata:
    id: GRAPH-EXAMPLE-001
    name:
      en: Example Data Product Graph
    description:
      en: A simple graph connecting a use case, data product, KPI, and objective.
  nodes:
    - id: UC-001
      type: UseCase
      $ref: ../usecases/customer-retention.yaml
    - id: DP-001
      type: DataProduct
      $ref: ../products/customer-360.yaml
    - id: KPI-001
      type: KPI
      $ref: ../kpis/churn-rate.yaml
    - id: OBJ-001
      type: BusinessObjective
      $ref: ../objectives/reduce-churn.yaml
  edges:
    - from: UC-001
      to: DP-001
      type: uses
      confidence: high
    - from: DP-001
      to: KPI-001
      type: tracks
      confidence: medium
    - from: KPI-001
      to: OBJ-001
      type: measures
      confidence: high

Testing

Run the Python toolkit test suite:

python -m pytest tests/test_odpg_toolkit.py

The tests cover validation, invalid edge handling, graph summaries, relationship traversal, strategic and governance analysis, AI-agent context extraction, and JSON output writing.

Docker

Build and run the documentation site with Docker:

docker build -t odpg-docs .
docker run --rm -p 4567:4567 odpg-docs serve

Build the static documentation output:

docker run --rm -v "${PWD}/build:/srv/slate/build" odpg-docs build

Relationship to the Open Data Products Family

Specification	Role
ODPS	Defines the structure and metadata of a data product
ODPV	Defines shared vocabulary and semantic meaning
ODPC	Defines catalog interoperability and discovery structures
ODPG	Defines graph relationships between data products, use cases, objectives, policies, agents, and related entities

ODPG should reference full data product metadata rather than copying it. Use ODPS for detailed product metadata, ODPV for stable vocabulary and semantics, ODPC for discovery catalogs, and ODPG for graph relationships, traversal, governance propagation, and strategic reasoning.

Contributing

Contributions are welcome through issues and pull requests. Useful contribution areas include:

• Specification clarifications and examples
• Schema improvements
• Toolkit enhancements
• Additional graph format converters
• Validation and analysis rules
• Documentation and visual explorer improvements

Please follow the repository code of conduct and keep changes focused, testable, and aligned with the ODPG graph model.

License

This project is licensed under the Apache License 2.0.

Development of the specification is under the umbrella of the Linux Foundation.