Salmon Data Integration System

This repository documents how the salmon ontology + data package tooling works together as one integration system.

Start here (overview + demo)

• System overview page: https://br-johnson.github.io/salmon-data-integration-system/
• Overview video (YouTube): https://youtu.be/B0Zqac49zng?si=VmOjbfMDMd2xW9fH
• Source repository: https://github.com/Br-Johnson/salmon-data-integration-system

Core repositories and tools

• DFO salmon ontology: https://github.com/dfo-pacific-science/dfo-salmon-ontology
• Salmon domain ontology (shared domain model): https://github.com/salmon-data-mobilization/salmon-domain-ontology
• Salmon domain ontology hub/docs: https://github.com/salmon-data-mobilization/salmon-ontology-hub
• Salmon Data Package (SDP) specification: https://github.com/dfo-pacific-science/smn-data-pkg
• metasalmon R package: https://github.com/dfo-pacific-science/metasalmon
• Salmon Data GPT app: https://chatgpt.com/g/g-69375eab4f608191863e8c23313a6f9f-salmon-data-standardizer

How the system fits together

Think of this as a layered workflow:

1. Shared domain semantics live in salmon-domain-ontology.
2. DFO-specific policy/operational semantics live in dfo-salmon-ontology.
3. Data exchange format is defined by the SDP spec (smn-data-pkg).
4. Operational data tooling in R is provided by metasalmon.
5. AI-assisted standardization workflow is supported by the Salmon Data GPT app.

In short: ontology layers define meaning, SDP defines structure, metasalmon handles validation/transformation in R, and the Salmon Data GPT app helps people standardize incoming datasets into that workflow faster.

Quick how-to

1) Start with ontology terms

• Identify the concepts and relationships your dataset needs.
• Reuse shared terms from salmon-domain-ontology where possible.
• Use DFO terms from dfo-salmon-ontology for DFO-specific context.

2) Use the Salmon Data GPT app to prepare data

• Open the GPT app and provide your dataset context (table structure, field meanings, units, source notes).
• Ask it to propose SDP-aligned field mappings and ontology term candidates.
• Use its output to produce a draft standardization plan/checklist before packaging.

3) Build a Salmon Data Package

• Follow the SDP spec in smn-data-pkg.
• Create package metadata and tables according to the required schema.
• Apply the reviewed mappings from Step 2.

4) Validate and work with packages in R

• Use metasalmon to inspect, validate, and transform package data.
• Resolve schema or semantic mapping issues before publishing/exchange.

5) Integrate downstream

• Publish/share valid SDPs.
• Use ontology mappings to support interoperability across teams, systems, and analyses.

Suggested implementation pattern

• Model first (ontology alignment)
• Standardize second (GPT-assisted mapping + preparation)
• Package third (SDP-compliant data product)
• Validate fourth (metasalmon checks)
• Integrate fifth (analytics, apps, services)

That order minimizes rework and keeps semantics + structure aligned from the start.

Important note

The GPT app is an accelerator, not the source of truth. Authoritative standards remain the ontologies + SDP spec, with final validation performed through reproducible package tooling (for example, metasalmon).