Building Shared Language for Salmon Knowledge

Building Shared Language for Salmon Knowledge Blueprint for a Salmon Knowledge Portal Nov 󳼔󳼔2󳼔󳼗, 󳼕󳼓󳼕󳼘 · Anchorage, Alaska Presented by: Research Data Alliance Salmon Ontology Working Group Authors: Brett Johnson (DFO) ·Tom Bird (DFO) ·Graeme Diack (AST) Shirly Stephens (NCEAS)·Melissa Morrison (DFO) Overview Language exists in contexts that express complex ideas. We aim to scale a method to express those contexts and properties explicitly Concepts with properties and relationships allow us to contextualize, AND share nuance between contexts Presentation Outline 1The Power and Peril of Words How inconsistent terminology creates barriers in salmon science and management 2Ontologies in Action How structured knowledge already powers systems we use daily 3The RDA Salmon Ontology Working Group Our collaborative approach to building shared language infrastructure 4Convergence Roadmap How DFO and NCEAS ontologies work together while respecting sovereignty 5Salmon Knowledge Portal Blueprint & Architecture From concept to reality - a blueprint for salmon knowledge exchange 6Call to Action How to get involved. Section 1: The Power and Peril of Words Language shapes how we understand salmon populations, make management decisions, and measure conservation success. Yet the same term can mean fundamentally different things across programs, jurisdictions, and communities. When definitions diverge, we risk more than confusion4we risk the fish and the communities that's depend on them. 2 Canadians, a French Guy, and a Scottish Guy were Sitting in a Pub& (Based on a True Story) Wild Salmon Meaning in Different Contexts Consider the term "Wild Salmon". A phrase word that fragments into distinct meanings depending on who's using it and why Depending on your context you might mean "wild salmon" when you use these terms: Contexts Hatchery Operations Adipose-Intact: Identified by fin clip status, distinguishing hatchery-reared from wild fish. Salmon Genetics Natural-Origin: Born in the wild without hatchery parentage, focusing on genetic lineage and reproductive origin. Stock Assessment Naturally Spawning: Reproducing in natural habitat, regardless of origin, Share Some Confusing Salmon Terms / Raise your Hand or Share in the Google Doc The Confusing Terms What specific salmon related words or definitions have you encountered that caused ambiguity or misunderstanding? The Root of the Confusion Describe why these terms led to confusion. Was it due to conflicting definitions, different contexts, or jargon? The Real-World Impacts What were the consequences of this terminological confusion? Did it affect decisions, communication, or conservation efforts? Impacts of Misalignment Miscommunication Delays Action Clarifying definitions consumes time better spent on conservation. Critical decisions are delayed while teams align on basic terms. Inconsistency Creates Waste Programs build unique glossaries and reconciliation scripts. This duplicated effort across organizations drains limited resources. Lack of Documentation Breeds Misunderstandings & Distrust Poorly documented data or implicit context makes scientists fear misinterpretation, leading to reluctance in sharing. Inconsistent or Colonial Terminology Reinforces Power Imbalances When technical jargon is used or certain groups control place names, it can exclude community voices and prevent equitable participation in salmon management decisions. Confusion Harms Fish Wrong habitat type, wrong season4an entire year of spawners lost because "juvenile rearing area" meant different things to different teams. How We Cope Today Current workarounds are resourceful but don't scale4they're patches, not infrastructure: Ad Hoc Glossaries & Data Dictionaries Email clarifications and meeting notes capture definitions, but they're siloed, version-controlled nowhere, and forgotten by the next project. Bespoke Crosswalks Every data integration spawns a new Excel spreadsheet mapping terms. These one-off translations don't accumulate into shared knowledge. Manual Reconciliation Scripts and manual reviews reconcile datasets case-bycase. Each synthesis project starts from scratch, reinventing the wheel. Extra Coordination Meetings Entire meetings dedicated to aligning on definitions before real work begins. Time lost that can't be recovered. These are patches, not infrastructure. We need a foundation that scales, persists, and serves the entire community. Fisheries Science Advice Ontology Conceptual Model Understanding the Model Our Conceptual Model illustrates the relationships between data sources, scientific processes, and decisionmaking pathways in fisheries management. It serves as a blueprint for tracing decisions back to data Improving Trust & Transparency The model ensures that scientific findings are transparently applied to critical fisheries decisions, from stock assessments to harvest control rules and conservation plans. What problem could an Ontology Solve for You? User Story: As a Salmon Biologist or decision-maker, I need to be able to refer to placenames by local, cultural and legal identifiers, so that I can communicate knowledge seamlessly in different contexts. User story as a Limmerick: A fisheries manager from the town of Torbay Scrunched up his map in dismay "There are 󳼘 'Muddy Brook's I can't find where to look, for the Eels I should be counting today." Section 3: Dissecting Salmon Terms Now it's your turn. Let's demystify ontology design by breaking down real terms together. Every shared understanding begins with careful examination of a single term. Let's practice that skill right now. The Problem with Compound Terms In salmon data, terms often seem straightforward, but phrases like "Natural Origin Spawner Abundance", "Commercial Catch", and "Smolt to Adult Return Rate" can be deceptively simple. These compound terms pack multiple concepts, methods, and assumptions into single phrases, often hiding critical context. Compound Term Hidden Context Implicit Assumptions Methodology Details Spatial & Temporal Scope Current metadata systems struggle to capture the granularity needed for these terms. Time Check < 󳼕󳼓 min skip Dissecting Salmon Terms "To address the fear of misunderstanding our data, we can breakdown terms to provide context" Let's break it down together "Mark3recapture escapement estimate for Cultus Lake sockeye, 2022" How could we break this down to clarify & feel confident others would understand the context and not misuse the data? "Petersen estimator with 󳼜󳼘% confidence intervals?" "Chapman modification for small sample bias?" "Wait... which tagging event and recovery period?" Step 1: Start with what's being measured (The Entity) "Mark3recapture escapement estimate for Cultus Lake sockeye, 2022" The "thing we're trying to measure": A fish? A Population? An ecological Process? Entity Population of Cultus Lake Sockeye Step 2: What characteristic is being measured "Mark3recapture escapement estimate for Cultus Lake sockeye, 2022" Entity Population of Cultus Lake Sockeye Characteristic Number of spawners Step 3: What method was used to measure the characteristic? "Mark3recapture escapement estimate for Cultus Lake sockeye, 2022" Describe how the characteristic was measured: the Mark2recapture protocol. Every scientific claim needs to show its methodological foundation. The protocol tells us HOW the estimate was derived. Entity Population of Cultus Lake Sockeye Characteristic Number of Spawners Measurement Method Mark-recapture (PIT tag? Floy? Disk???) Myth vs. Fact  Myth "Ontologies are mostly academic exercises with little practical value." 7 Fact "Ontologies power systems you use daily4 Google Maps, NASA data, and e-commerce recommendations all rely on structured knowledge."  Myth "Building ontologies requires advanced technical skills and years of training." 7 Fact "Our learning pathway takes you from vocabulary reuse to ontology creation step-bystep, starting with simple term documentation." You maintain control over your terms, definitions, and context. The ontology provides the reference layer to connect your knowledge to others4when and how you choose. Section 4. Converging Streams of Knowledge: Roadmap Aligning efforts, organizing conventions, integrating knowledge. Review existing ontology efforts Discuss integration and federation strategies Roadmap for Ontology Development & Knowledge Portal Presentation Outline 1The Power and Peril of Words 2Ontologies in Action 3Dissecting Salmon Terms 4The RDA Salmon Ontology Working Group 5Convergence Roadmap 6Salmon Knowledge Portal Blueprint & Architecture NCEAS Salmon Ontology on BioPortal The NCEAS Salmon Ontology was developed as part of the 'State of Alaska Salmon and People' (SASAP) initiative and provides a structured and standardized vocabulary for salmon research. Published on BioPortal, a widely recognized repository for biomedical ontologies, this resource has the potential to serve a 'domain ontology' facilitating broad accessibility and interoperability of salmon data. Umbrella Model: Domain Interoperability Reference Layer NCEAS provides the shared umbrella that enables different organizations to connect while maintaining their unique identities and local control. NCEAS Salmon Domain Ontology DFO Salmon Ontology Canadian Gov't Specific Terms SASAP Ontology Alaska-focused data and context [Your Context Here] Integrate additional data sources NCEAS Salmon Ontology (The Umbrella) Role: Domain/interoperability layer covering cross-agency salmon biology, ecology, and fisheries concepts. Examples: Species classifications, habitat types, common fishery categories, standard measurements. Function: Provides shared semantics that everyone can align to. Under the Umbrella - Organizational Ontologies DFO Salmon Ontology: Canadian federal management and assessment context. SASAP (State of Alaska Salmon and People): Alaska state programs and community knowledge. [Your Org/Project/Application Here]: Placeholder for any organization or project. Core Principle: Each organization keeps their local detail, context, and control while aligning to the NCEAS umbrella for shared domain concepts. This enables interoperability without sacrificing organizational specificity. Federated queries across ontologies and knowledge graphs could underpin the Salmon Knowledge Portal Roadmap for NCEAS Salmon Domain Ontology This strategic roadmap outlines key initiatives to establish the NCEAS Salmon Ontology as the definitive domain ontology, built on collaborative development and iterative releases. 1 Community Review & Feedback Engage diverse stakeholders to solicit feedback on initial drafts and ensure broad applicability across regions and disciplines. 2Establish Conventions Define clear modeling patterns, naming conventions, and best practices to promote consistency and future extensibility. 3 Contributing Guidelines Develop comprehensive instructions and streamlined workflows to empower external experts to enrich the ontology sustainably. 4Refactor for Interoperability Integrate feedback, expand domain-wide terminology, and optimize the ontology for crosssystem data exchange. 5 v0.1.0 Release Launch the initial stable version, providing a foundational resource for early adopters and ongoing collaborative development. 6Start Accepting New Terms from Community Contributions 7 Leverage Federated Salmon Ontologies for Salmon Knowledge Portal Section 5. Realizing the Knowledge Portal Vision The ontology work connects directly to a tangible product: the Salmon Knowledge Portal4a unified interface for discovering, exploring, and synthesizing salmon data across systems. Knowledge Portal Components The portal is what users see; the ontology translates between interface and data systems. Search & Discovery Interface Plain-language queries, map-based exploration, narrative storytelling, and visualization tools accessible to diverse audiences. Salmon Ontology (Translation Layer) The ontology acts as a common language, allowing data to remain in its original local formats and terminologies while being mapped or linked to standardized terms. Distributed Data/Information Systems DFO databases, NCEAS repositories, tribal monitoring programs, State databases, research archives4each maintaining sovereignty and control. Data stays where it lives. The ontology creates a shared language that enables federation without centralization. Salmon Knowledge Portal Blueprint The technical blueprint for the Salmon Knowledge Portal, integrating user interface, data, governance, and learning pathways. Sociocultural Dimensions: Local language and context remains preserved, and mapped to the ontology Move us beyond tabular data Data Systems: Distributed architecture respecting sovereignty Independent catalogues using shared technology s they can be ingested into Portal User Experience Simple input options, plain language queries, story-maps, visualizations, info sheets, data products Semantic Infrastructure: NCEAS Salmon Ontology as our domain reference Alignment patterns for interoperability API & Integration: API Endpoints, SPARQL Query Engine Schema.org markup Data Exchange & Federation Protocols From Vision to Blueprint Last year, your feedback provided a clear vision. This year, we've developed an actionable blueprint. The Shift: From inspiring vision to actionable blueprint. From community needs to technical specifications. From "we should" to "here's how." 2024 Outcomes ³ 2025 Responses 2024 Outcomes Data ³ Knowledge Indigenous inclusion Project visibility & funding Neutral steward Decision context 2025 Responses Portal frames knowledge (products, plain language, visuals), not just data Domain driven design; sovereignty & consent encoded; story-map modeling Discovery-level metadata + funding linkages in the ontology Governance-first approach; multilateral oversight before production These responses are embedded in the architecture, governance model, and training materials, moving us from concept to implementation. What would you ask the Salmon Knowledge Portal? Imagine the portal is live right now. You can type a question in plain language and get an answer that synthesizes data across agencies, decades, and geographies. Prompts to consider: What decision are you trying to inform? What comparison would reveal something important? What pattern would change your understanding? What question have you wanted to ask but couldn't because the data was too fragmented? Your questions will directly shape the competency questions we use to design and test the ontology. This is your chance to influence what the portal becomes RDA Ontology Development Timeline 󳼔󳼛-month roadmap with clear milestones, collaboration rhythms, and open outputs: 1 Jun-July 2025: Discovery Use-case brief, module requirements, initial dataset modelling 2Aug-Sept 2025: Case Study Integration demo and workshop outline 3 Oct 2025-Feb 2026 Alpha Materials Draft training materials and conventions review 4March 2026: Draft Recommendation Carpentry-style materials ready for pilot testing 5 April 2026: Pilot Workshop Live training with real participants 6December 2026: Final Recommendation Published workshop materials on GitHub/Zenodo under CC BY 󳼗.󳼓 Tools & Standards SKOS, OWL, ENVO/OBO ROBOT, WebProtégé, Large Language Models GitHub Coordination RDA Salmon R&M IG, Vocabulary Services IG, I-ADOPT, ESIP STC (observer) Meeting Cadence: Full WG every 󳼙 weeks; task teams biweekly (Discord) Integration Strategy: Align, Contribute, Don't Duplicate How modules plug into the domain umbrella4balancing reuse with local control: Reuse NCEAS Classes For shared concepts, use subClassOf or equivalence where semantically valid Keep Org-Specific Local Maintain DFO namespace for policy-specific or jargon-heavy concepts where precision matters Contribute Back General methods (AUC, mark-recapture, weir counts) become SKOS/OWL contributions in the domain ontology so everyone benefits Share Vocabularies Joint schemes for methods and types; link via skos:exactMatch or closeMatch dfo:PreTerminalFishery rdfs:subClassOf nceas:FisheryType skos:Concept dfo:AUC_Method skos:exactMatch nceas:AreaUnderCurve Example Alignment What We Won't Do Overwrite local meanings Erase context or nuance Centralize control