ontinkering

Ontinkering is a term used in knowledge engineering and semantic web communities to describe the iterative, hands-on development and refinement of ontologies. It emphasizes practical experimentation with concepts, classifications, relationships, and constraints to better reflect domain knowledge and data interoperability. The approach favors exploration, prototyping, and incremental changes guided by real-world data and use cases over formal, document-driven methods alone.

The word is a portmanteau of ontology and tinkering. It appears in discussions among knowledge engineers, data scientists, and AI researchers focusing on knowledge graphs and linked data. While not always rigorously defined, ontinkering signals a mindset that values learning-by-doing, rapid feedback, and continuous improvement of knowledge structures as data ecosystems evolve.

Typical activities include collecting competency questions, testing ontologies with reasoners, adding classes or properties, restructuring hierarchies, and iteratively aligning with external vocabularies or upper ontologies. Toolchains often involve version control, collaborative editing, automated checks, and small-scale demonstrations that reveal gaps or ambiguities in definitions and constraints.

Applications span domains such as biomedical informatics, e-commerce catalogs, cultural heritage metadata, and enterprise data integration, where evolving terminology and diverse data sources demand flexible, incremental modeling. Critics note that ontinkering can lead to less formal documentation, potential inconsistencies if changes are not coordinated, and governance challenges for long-term maintenance.

As a practice, ontinkering sits alongside more formal ontology engineering traditions, offering a pragmatic approach to keeping knowledge representations usable and aligned with current data realities.