Date of Award

12-14-2017

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Science

First Advisor

Veda C. Storey

Second Advisor

Raj Sunderraman

Third Advisor

Bala Ramesh

Fourth Advisor

Vijayan Sugumaran

Abstract

As the number of intelligent software applications and the number of semantic websites continue to expand, ontologies are needed to formalize shared terms. Often it is necessary to either find a previously used ontology for a particular purpose, or to develop a new one to meet a specific need. Because of the challenge involved in creating a new ontology from scratch, the latter option is often preferable. The ability of a user to select an appropriate, high-quality domain ontology from a set of available options would be most useful in knowledge engineering and in developing intelligent applications. Being able to assess an ontology's quality and suitability is also important when an ontology is developed from the beginning. These capabilities, however, require good quality assessment mechanisms as well as automated support when there are a large number of ontologies from which to make a selection.

This thesis provides an in-depth analysis of the current research in domain ontology evaluation, including the development of a taxonomy to categorize the numerous directions the research has taken. Based on the lessons learned from the literature review, an approach to the automatic assessment of domain ontologies is selected and a suite of ontology quality assessment metrics grounded in semiotic theory is presented. The metrics are implemented in a Domain Ontology Rating System (DoORS), which is made available as an open source web application. An additional framework is developed that would incorporate this rating system as part of a larger system to find ontology libraries on the web, retrieve ontologies from them, and assess them to select the best ontology for a particular task. An empirical evaluation in four phases shows the usefulness of the work, including a more stringent evaluation of the metrics that assess how well an ontology fits its domain and how well an ontology is regarded within its community of users.

Share

COinS