Please use this identifier to cite or link to this item:
http://hdl.handle.net/11375/27280
Title: | A Formal Approach to Ontology Modularization and to the Assessment of its Related Knowledge Transformation |
Authors: | LeClair, Andrew |
Advisor: | Khedri, Ridha |
Department: | Computing and Software |
Publication Date: | 2021 |
Abstract: | Knowledge-based systems are composed of multiple agents that each must utilize their understanding of the world to deduce new facts and make intelligent decisions. The understanding of the world that each agent has is formalized using an ontology: a structure which conceptualizes the domain as a set of concepts and their relations. To accommodate the numerous agents, and to minimize the cost of a single monolithic ontology, modularization is used to provide each agent with only the knowledge they require. This thesis addresses the problems related to defining the transformation of the domain knowledge due to ontology modularization. Existing ontology modularization techniques are evaluated so that they can be assessed with respect to how they transform knowledge, as well as their computational performances. A primary objective is to adapt the existing modularization techniques so that they can be compared based on how they transform domain knowledge. The objectives of this thesis were addressed systematically by first comprehensively evaluating the literature regarding both ontologies and ontology modularization. Then, we adapted the prominent ontology modularization techniques to the ontology formalism used in this thesis: a Domain Information System. New modularization techniques were then explored which utilized aspects of the domain ontology's algebra. This research consolidates the many approaches to ontology modularization into a single set of formal techniques for a domain ontology. As a result, we further the modularization field by unifying the modularization field under a common formalism, creating relations between the modularization techniques, as well as formulating how the knowledge of the ontology is transformed for each technique. Additionally, we introduce a new modularization technique which capitalizes on the Boolean algebra of a domain ontology, and enables the determination of a module based on the desired granularity of the domain knowledge under consideration. The research of this thesis furthers the ontology modularization field by systematically analyzing current limitations with respect to their ability to determine how knowledge is transformed. We demonstrate that by using the formalism of the domain ontology, we are able to formulate different modularization techniques, each with its own transformations to the domain knowledge. Expressing the knowledge that is transformed via modularization is unique. It is also instrumental in providing the agents of a knowledge-based system with modules that capture the exact domain knowledge they require to make their decisions. |
URI: | http://hdl.handle.net/11375/27280 |
Appears in Collections: | Open Access Dissertations and Theses |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
LeClair_Andrew_L_202112_phd.pdf | 908.35 kB | Adobe PDF | View/Open |
Items in MacSphere are protected by copyright, with all rights reserved, unless otherwise indicated.