Evaluation
Evaluation is a key component of ontology learning. Evaluation is typically performed by a human expert or knowledge
engineer; however, it is important that standards and guidelines for ontology evaluation are available and utilized. Precision
and recall are popular measures for evaluating ontology learning (Spyns & Reinberger, 2005). They are also referred to as
lexical precision and recall and the latter reflects how well the learned terms cover the domain (Dellschaft & Staab, 2006).
Similarly taxonomic precision and recall may be employed in the evaluation process. Taxonomic recall reflects how well the
structure – or taxonomy – reflects the domain. It is also possible to conduct evaluations based on reference ontology
(Dellschaft & Staab, 2006). In this scenario, a human expert creates domain ontology manually. The learned ontology is then
compared with the manual ontology to determine the accuracy of the learned ontology. Ontologies may also be compared on
instance data or on the schema. Instance based evaluation is concerned with the instances and classification of the instances,
whereas schema based evaluation compares the ontologies at the schema level.