Dataset: Evolution Of Computational Ontologies: Assessing Development Processes Using Metrics

Ontologies facilitate meaning between human and computational actors. On the one hand, the underlying technology can be considered mature. It has a standardized language, established tools for editing and sharing, and broad adoption in practice and research. On the other hand, we still know little about how these artifacts evolve over their lifetime, even though knowledge of the development process could influence quality control. It would enable us to give knowledge engineers better modeling or selection guidelines. This paper examines the evolution of computational ontologies using ontology metrics. First, we gathered hypotheses on the ontology development process. We assume that groups of ontologies follow a similar development pattern and that a stereotypical development process exists. Afterward, these hypotheses are tested against historical metric data from 7053 versions from 69 dormant ontologies. We will show that ontology development processes are highly heterogeneous. While the made hypotheses are partly true for a slight majority of ontologies, concluding the bigger picture of ontology development down to the individual ontologies is mostly not possible. Further, the data revealed that most ontologies have disruptive change events for most of the measures attributes. These disruptive events are further examined regarding their occurrences, combinations, and sizes.