Salmon Data Integration System

• Project page: br-johnson.github.io/salmon-data-integration-system
• Overview video: youtu.be/B0Zqac49zng
• Source repository: github.com/Br-Johnson/salmon-data-integration-system

• DFO salmon ontology: github.com/dfo-pacific-science/dfo-salmon-ontology
• Salmon domain ontology (shared domain model): github.com/salmon-data-mobilization/salmon-domain-ontology
• Salmon domain ontology hub/docs: github.com/salmon-data-mobilization/salmon-ontology-hub
• Salmon Data Package (SDP) specification: github.com/dfo-pacific-science/smn-data-pkg
• metasalmon R package: github.com/dfo-pacific-science/metasalmon
• Salmon Data GPT app: chatgpt.com/g/.../salmon-data-standardizer

1. Shared domain semantics live in salmon-domain-ontology.
2. DFO-specific semantics live in dfo-salmon-ontology.
3. Data exchange structure is defined by the SDP spec (smn-data-pkg).
4. R-based validation and transformation are handled by metasalmon.
5. AI-assisted standardization is accelerated by the Salmon Data GPT app.

1) Start with ontology alignment

• Identify dataset concepts, entities, and relationships.
• Reuse shared terms from salmon-domain-ontology where possible.
• Use DFO terms from dfo-salmon-ontology for policy/operational context.

2) Use the Salmon Data GPT app

• Provide dataset context (columns, units, source notes, known constraints).
• Ask for SDP-aligned field mapping suggestions and ontology term candidates.
• Use the generated plan as a draft, then review it before final packaging.

3) Build an SDP-compliant package

• Follow the specification in smn-data-pkg.
• Create package metadata and required data tables.
• Apply reviewed mappings from the GPT-assisted standardization step.

4) Validate with metasalmon

• Inspect and validate structure + content in R.
• Fix schema and semantic issues before sharing or publishing.

5) Integrate downstream

• Share valid SDPs across teams/systems.
• Leverage ontology mappings to maintain interoperability for analytics and apps.