Automatic & Semi-Automatic Methods for Supporting Ontology Change

The Semantic Web vision consists in encoding the knowledge made available on the web in a way which can be understood by machines as well as by humans. Over the last years more and more semantic data has been published on the web. Semantic data is made available in the form of ontologies – formal models of a domain of interest. As the underlying domain or the requirements for a specific ontology may change, the knowledge encoded in it is also changing over time. Therefore, changes to the ontologies are necessary. However, these changes are complicated, as the knowledge in the ontology is usually highly interlinked and only experts may be able to understand the complete formalization. The main topic of this thesis is the development of methods which support and thus facilitate the ontology change process. The contributions of this thesis are threefold: Firstly, existing change support methods are analysed and classified according to a newly proposed classification scheme. Secondly, a method for inducing additional facts from the existing ontology is proposed. Our contribution lies in the definition of kernel functions for Resource Description Framework (RDF) data which may be used for classifying entities as well as for predicting links between entities. The proposed methods may be used for completing an existing ontology with facts that have not explicitely been stated and are not derivable (or rejectable) by means of deductive methods. Thirdly, a framework is proposed which allows for the automatic handling of complex update requests. The approach is based on the identification of change patterns, which describe sets of frequently occurring changes. Such changes may be predefined in the proposed framework and then be instantiated later on. This approach allows users who are not familiar with the precise formalization of knowledge in the ontology to perform changes on it.