Potentials and Limits of Graph-Algorithms for Discovering Ontology Patterns

The aim of the research presented in this paper is to contribute to reduction of development time for ontologies by discovering reusable parts and patterns. The approach is to use existing theories and algorithms from graph theory in order to investigate the possibility of graph-pattern reoccurrence in ontologies. The experiments performed are based on a definition of what structures are considered patterns and use the tool Subdue for exact and inexact matching, applied on a set of ontologies. Increasing the threshold for variation in inexact matching leads to more interesting patterns, but the precision decreases significantly and drops quite fast to less than 25%. The main conclusion is that graph-pattern algorithms appear inefficient for finding patterns in ontologies, due to the more or less infinite different ways of assigning a name to a concept or relation and structuring the ontology. In order to use graph theories for finding patterns, significant improvements to the pattern identification process have to be made.