An Approach to Produce a Quality Ontology for a Usable Ontology-driven Information System

Abstract

Abstract Nowadays, ontologies are increasingly incorporated in information systems owning to the distinct capabilities that they have. Most specifically, ontologies have the significant feature of inferencing knowledge from explicit facts which is not available in other repositories and data models. However, these capabilities would not be beneficial for Ontology-Driven Information Systems (ODIS) unless a good quality ontology is produced. This is due to the fact that ontology quality strongly affects the overall system quality and in turn, user satisfaction. This situation has been realized in several previous works, as well as, through this study. For instance, the studies show that even very few quality issues in an ontology can lead to producing substandard results for user queries. Consequently, this effect creates a negative user attitude toward the system adversely affecting its usability in use. Accordingly, it is clearly evident that a good quality ontology is necessary to make the overall ODIS usable. Thus, a proper quality assessment of ontology content and structure is essential before it is deployed in a system. However, as realized through the literature, the existing works have not thoroughly considered the quality requirements and their evaluation across ontology development. The main cause for this issue is the non-existence of agreed models, methodologies and approaches which support researchers and developers to specify and evaluate the quality characteristics of an ontology. This dissertation alleviated the aforementioned problem by introducing an ontology quality model and approach. Initially, a systematic review was performed to identify the existing contribution to the quality evaluation, i.e., quality characteristics, measures and methods. Consequently, a model was constructed with nineteen characteristics classified into four evaluation scopes namely structural intrinsic, domain intrinsic, domain extrinsic and application extrinsic. Thereafter, an approach was built by adopting well-formed theories and standards in software quality engineering. The overall research study was undertaken by following the Design Science Research methodology. The evaluation of the quality model was performed with the assistance of specialists in ontology quality and the evaluation of the quality approach was carried out using a use case in the agriculture domain. The results show that the proposed model and approach are successful artifacts that support researchers and practitioners to specify and evaluate the quality of an ontology. In the future, it is expected that the developed model and approach will be applied to many case studies in different domains to make them more complete and well-established.