Computational ontologies are key to information retrieval, semantic integration of datasets, and semantic similarity analyses. It's sometimes a challenge to find a self-contained material where you can find "what" to evaluate after you've designed an ontology (criteria), and "how" to do it (strategies). Here you may find a start. Depending on the use case, and the intended audience, it may be interesting to evaluate design aspects, or implementation aspects, or both.
-
Accuracy (correct representation of aspects of the real world)
-
Adaptability (ease of performing changes)
-
Clarity (effective communication of the intended meaning of defined terms)
-
Cognitive adequacy (match between formal and cognitive semantics)
-
Completeness (appropriate coverage of the domain of interest)
-
Conciseness (absence of unnecessary or useless definitions or axioms)
-
Consistency (incapacity of getting contradictory conclusions from valid input data)
-
Expressiveness (number of competency questions that the ontology can answer)
-
Grounding (number of assumptions done by the ontology’s underlying philosophical theory about reality)
-
Computational efficiency (ease and speed of processing by reasoners)
-
Congruency (fitness between ontology and corpus terms)
-
Practical usefulness (number of practical problems to which the ontology can be applied)
-
Precision (fraction of retrieved instances by the ontology that are relevant)
-
Recall (fraction of relevant instances that are retrieved by the ontology)
The list will be extended with strategies as I become aware of them. Reference: Degbelo, A. (2017). A snapshot of ontology evaluation criteria and strategies. In R. Hoestra, C. Faron-Zucker, T. Pellegrini, & V. de Boer (Eds.), Proceedings of the 13th International Conference on Semantic Systems - SEMANTICS 2017 (pp. 1–8). https://doi.org/10.1145/3132218.3132219