Service Level Agreement Template

&

Annotation Acceptance Framework

Roadmaps

​

Wouter Addink

​

National Nodes meeting and ISTC meeting, May 2026

license: CC-By

​

Service Level Agreement Template

​

Purpose: prepare for SLAs (Service Level Agreements) that will be used as the formal contractual agreements between a Service Provider and DiSSCo ERIC.

​

Scope: Both core and non-core DiSSCo services

​

Proposed solution: One or more SLA templates with the Service Provision General Framework Agreement (SP/GFA) as foundation (https://doi.org/10.5281/zenodo.17701077)

​

​

​

SLA Template status

​

• 0.1 version draft available: https://bit.ly/DiSSCo_SLA
• First public consultation meeting held 18 May
• Feedback requested by email

​

​

SLA template development Roadmap

Aim:a finalized plan by the end of 2026

​

Annotation Acceptance Framework (AAF)

​

Purpose: to govern the submission, validation, and acceptance (or rejection) of annotations

​

Scope: Annotations on Digital Specimen and Digital Media objects

​

Proposed solution: custodian can transfer responsibility to experts, separation in high and low impact changes.

​

​

TAB Recommendation: "Layered Governance" Model

The Technical Advisory Board (TAB) strongly recommended moving away from a rigid, "gatekeeper" style of data management toward a democratized, layered approach.

• Move Beyond the "Single Point of Truth": Instead of forcing a single version of a record, DiSSCo should maintain a Knowledge Graph where all annotations coexist as linked data.
• The "Community Consensus View": DiSSCo should define a "Blessed" or "Primary" subgraph. This is the version of the data that has been vetted by chosen experts or trusted institutions.
• User Choice & Transparency: End-users and external organizations (like GBIF) should be able to choose which "layer" of trust they want to view—whether it’s the expert-vetted data, the raw community discussion, or a specific institution’s preferred records.
• Avoid Bespoke Silos: Stick to multi-domain open standards (e.g., Linked Open Data, FDOs) rather than building technical solutions tailored only to specific stakeholders.

Technical Solution: Framework Architecture�Goal: balance automation for routine tasks with expert oversight for critical scientific changes.

A. Categorization of Changes by Scientific Impact

• Low-Impact (Automated): Simple error corrections (e.g., longitude/latitude formatting or obvious typos). These can be accepted automatically via trusted machine services.
• High-Impact (Vetted): Changes to scientific identification (e.g., a new species name). These require human expert or "Trusted Machine" verification.

B. The Expert Layer

• Expert Groups: Recognition of community-driven expert groups who have the authority to vet annotations.
• Machine Annotation Services: AI or automated tools can be granted "expert" status if the community trusts their logic, allowing them to handle high volumes of routine data cleaning.

C. Provenance & Versioning

• Provenance: Every annotation is a digital object with clear provenance
• Versioning: DOIs point to latest state of the annotated object but individual versions remain citable

Core Principles for Implementation

• Provenance and Persistence (Technical Requirement)
• Evidence and Justification (Scientific Requirement)
• Right to revert changes (Custodian Requirement)
• Conflict Resolution (Governance Requirement)

To achieve layered governance:�Data Custodian can delegate responsibility to an Expert(-group), Citizen Scientist or MAS

AAF development Roadmap

Aim:a finalized plan by the end of 2026