Datakit is thinking about how to solve the problem of establishing a reliable means of optimizing data for long-term use. They feel existing approaches doen't work:
- STEP has limits that are liable to impede its extension going forward.
- Vendor-defined formats have gray areas, such as how to deal with annotations, symbols, features, kinematics, 2D, and management data.
Instead, Datakit proposes a method for describing and classifying data, and using specific access and query methods. They call it the "ontological approach":
An ontology is a structured set of concepts that gives meaning to information. It is a basis for formalizing knowledge that classifies relations between concepts into categories.