As information systems become more complex it is infeasible for a non-expert to understand how the information system has evolved. Accurate models of these systems and the changes occurring to them are required for interpreters to understand, reason over, and learn from evolution of these systems. Ontologies purport to model the semantics of the domain encapsulated in the system. Existing approaches to using ontologies do not capture the rationale for change but instead focus on the direct differences between one version of a model and the subsequent version. Some changes to ontologies are caused by a larger context or goal that is temporally separated from each specific change to the ontology. Current approaches to supporting change in ontologies are insufficient for reasoning over changes and allow changes that lead to inconsistent ontologies. / In this thesis we examine the existing approaches and their limitations and present a four-level classification system for models representing change. We address the shortcomings in current techniques by introducing a new approach, augmenting ontologies with operations for capturing and representing change. In this approach changes are represented as a series of connected, related and non-sequential smaller changes. The new approach improves on existing approaches by capturing root causes of change, by representing causal relationships between changes linking temporally disconnected changes to a root cause and by preventing inconsistencies in the evolution of the ontology. The new approach also explicitly links changes in an ontology to the motivating real-world changes. We present an abstract machine that defines the execution of operations on ontologies. A case study is then used to explain the new approach and to demonstrate how it improves on existing ways of supporting change in ontologies. The new approach is an important step towards providing ontologies with the capacity to go beyond representing an aspect of a domain to include ways in which that representation can change.

As information systems become more complex it is infeasible for a non-expert to understand how the information system has evolved. Accurate models of these systems and the changes occurring to them are required for interpreters to understand, reason over, and learn from evolution of these systems. Ontologies purport to model the semantics of the domain encapsulated in the system. Existing approaches to using ontologies do not capture the rationale for change but instead focus on the direct differences between one version of a model and the subsequent version. Some changes to ontologies are caused by a larger context or goal that is temporally separated from each specific change to the ontology. Current approaches to supporting change in ontologies are insufficient for reasoning over changes and allow changes that lead to inconsistent ontologies. / In this thesis we examine the existing approaches and their limitations and present a four-level classification system for models representing change. We address the shortcomings in current techniques by introducing a new approach, augmenting ontologies with operations for capturing and representing change. In this approach changes are represented as a series of connected, related and non-sequential smaller changes. The new approach improves on existing approaches by capturing root causes of change, by representing causal relationships between changes linking temporally disconnected changes to a root cause and by preventing inconsistencies in the evolution of the ontology. The new approach also explicitly links changes in an ontology to the motivating real-world changes. We present an abstract machine that defines the execution of operations on ontologies. A case study is then used to explain the new approach and to demonstrate how it improves on existing ways of supporting change in ontologies. The new approach is an important step towards providing ontologies with the capacity to go beyond representing an aspect of a domain to include ways in which that representation can change.

As information systems become more complex it is infeasible for a non-expert to understand how the information system has evolved. Accurate models of these systems and the changes occurring to them are required for interpreters to understand, reason over, and learn from evolution of these systems. Ontologies purport to model the semantics of the domain encapsulated in the system. Existing approaches to using ontologies do not capture the rationale for change but instead focus on the direct differences between one version of a model and the subsequent version. Some changes to ontologies are caused by a larger context or goal that is temporally separated from each specific change to the ontology. Current approaches to supporting change in ontologies are insufficient for reasoning over changes and allow changes that lead to inconsistent ontologies. / In this thesis we examine the existing approaches and their limitations and present a four-level classification system for models representing change. We address the shortcomings in current techniques by introducing a new approach, augmenting ontologies with operations for capturing and representing change. In this approach changes are represented as a series of connected, related and non-sequential smaller changes. The new approach improves on existing approaches by capturing root causes of change, by representing causal relationships between changes linking temporally disconnected changes to a root cause and by preventing inconsistencies in the evolution of the ontology. The new approach also explicitly links changes in an ontology to the motivating real-world changes. We present an abstract machine that defines the execution of operations on ontologies. A case study is then used to explain the new approach and to demonstrate how it improves on existing ways of supporting change in ontologies. The new approach is an important step towards providing ontologies with the capacity to go beyond representing an aspect of a domain to include ways in which that representation can change.