Ontology evaluation approaches: A case study from agriculture domain

Abstract

The quality of an ontology very much depends on its validity. Therefore, ontology validation and evaluation is very important task. However, according to the current literature, there is no agreed method or approach to evaluate an ontology. The choice of a suitable approach very much depends on the purpose of validation or evaluation, the application in which the ontology is to be used, and on what aspect of the ontology we are trying to validate or evaluate. We have developed large user centered ontology to represent agricultural information and relevant knowledge in user context for Sri Lankan farmers. In this paper, we described the validation and evaluation procedures we applied to verify the content and examine the applicability of the developed ontology. We obtained expert suggestions and assessments for the criteria used to develop the ontology as well as to obtain user feedback especially from the farmers to measure the ontological commitment. Delphi Method, Modified Delphi Method and OOPS! Web-based tool were used to validate the ontology in terms of accuracy and quality. The implemented ontology is evaluated internally and externally to identify the deficiencies of the artifact in use. An online knowledge base with a SPARQL endpoint was created to share and reuse the domain knowledge. It was also made use of for the evaluation process. A mobile-based application is developed to check user satisfaction on the knowledge provided by the ontology. Since there is no single best or preferred method for ontology evaluation we reviewed various approaches used to evaluate the ontology and finally identified classification for ontology evaluation approaches based on our work.