Modelling and Use of Domain-Specific Knowledge for Similarity and Visualization
Henrik Bulskov, Rasmus Knappe, Troels Andreasen · RUCforsk (Roskilde University) · 2005
The use of ontologies can contribute significantly to the organization of concepts, structure and relations within a knowledge domain. Incorporation of ontologies in tools for information access provides foundation for enhanced, knowledge-based approaches to surveying, indexing and querying of document collections. We introduce in this paper the notion of an instantiated ontology as a subontology derived from a general ontology and restricted by the set of instantiated concepts in a target document collection. This instantiated ontology represents a conceptual organization reflecting the document collection, and reveals domain knowledge, for instance about the thematic areas of the domain which in turn facilitates means for a topic-based navigation and visualization of the structure within the domain. Modeling and use of ontologies is the major focus in this paper. We introduce, in section 2, to a formalism for representation of ontologies. Section 3 describes the modeling of general and instantiated ontologies respectively. In section 4 it is discussed how to establish a measure of domain-specific similarity from a so-called domain-specific ontology. Section 5 introduces the use of an instantiated ontology for domain and query visualization. Concept expressions, that are the key to modelling and use of ontologies, are explained in more detail below, we refer, however, to (Andreasen et al., 2004b) for a discussion of general principles behind parsing for concepts.