LinkML: an open data modeling framework

Area of Science:

• Data science
• Computational biology
• Information management

Background:

• Scientific data is often unstructured (e.g., free-text notebooks, spreadsheets), hindering interoperability.
• Lack of data standardization complicates integration, validation, and reuse.
• Existing data models can be complex and lead to proliferation of single-use formats.

Purpose of the Study:

• Introduce LinkML (Linked Data Modeling Language) as a solution for data standardization.
• Demonstrate LinkML's capability to simplify data authoring, validation, and sharing.
• Promote adoption of LinkML for enhanced data interoperability and FAIR compliance.

Main Methods:

• Utilizes an open framework with an approachable syntax for describing data schemas, classes, and relationships.
• Supports diverse data structures from simple lists to complex, normalized models with inheritance.
• Allows seamless integration with existing frameworks and import of schemas.

Main Results:

• LinkML enables the creation of well-defined, stable, and ontology-aligned data structures.
• Facilitates interdisciplinary collaboration through accessible data semantics.
• Simplifies the process of defining and sharing data models.

Conclusions:

• LinkML reduces data heterogeneity and complexity, promoting FAIR data standards.
• It has broad adoption across various scientific and commercial domains.
• LinkML makes implicit data models explicit and computable, standardizing data at its source.