Modellregistre

Modellregistre is a registry or repository that catalogs model definitions, configurations, and related metadata to support governance, reuse, and traceability across projects. It acts as a centralized source of truth for models used in areas such as product development, simulation, and data science, enabling versioning, provenance, and controlled access.

In practice, organisations use modellregistre as part of broader ecosystems such as product lifecycle management, machine learning operations, or data catalog systems. The register typically covers both technical models—such as CAD models, simulation models, and trained machine learning models—and other formal models used in planning, forecasting, or decision making. The objective is to provide clear ownership, reproducibility, and a record of decisions and changes.

A modellregistre usually stores metadata and attributes for each model, such as model identifier, name, version, type, description, owner, status or stage, approval date, deployment target, dependencies, training data and parameters, evaluation metrics, provenance, lineage, license, and access controls. It also maintains an audit log and links to related artifacts like datasets, code, tests, and documentation. This enables traceability from model creation to deployment and retirement.

Governance and processes are central to a modellregistre. Workflows typically include model creation, validation, approval, publishing, versioning, deployment, monitoring, and retirement. Roles such as model owner, reviewer, approver, and auditor are defined, along with access policies and retention rules. Interoperability with other systems and adherence to metadata standards are common priorities to ensure data quality and reuse.

Overall, a modellregistre supports transparency, compliance, and efficient management of models across an organization. The concept is often translated as model registry or model catalog in English.