Operational change management and change pattern identification for ontology evolution

by Muhammad Javed

Institution: Dublin City University
Department: School of Computing
Year: 2013
Keywords: Information technology; Software engineering; World Wide Web; semantic ontology evolution; ontology change patterns; pattern-based ontology evolution; change log graph; graph-based composite change detection; change pattern discovery algorithms
Record ID: 1182744
Full text PDF: http://doras.dcu.ie/18187/


Ontologies can support a variety of purposes, ranging from capturing the conceptual knowledge to the organization of digital content and information. However, information systems are always subject to change and ontology change management can pose challenges. In this sense, the application and representation of ontology changes in terms of higher-level change operations can describe more meaningful semantics behind the applied change. We propose a four phase process that covers the operationalization, representation and detection of higher-level changes in ontology evolution life cycle. We present different levels of change operators based on the granularity and domain-specificity of changes. The first layer is based on generic atomic level change operators, whereas the next two layers are user-defined (generic/domain-specific) change patterns. We introduce the layered change logs for an explicit and complete operational representation of ontology changes. The layered change log model has been used to achieve two purposes, i.e. recording of ontology changes and mining of implicit knowledge such as intent of change, change patterns etc. We formalize the change log using a graph-based approach. We introduce a technique to identify composite changes that not only assist in formulating ontology change log data in a more concise manner, but also help in realizing the semantics and intent behind any applied change. Furthermore, we discover the reusable ordered/unordered domain-specific change patterns. We describe the pattern mining algorithms and evaluate their performance.