Predicting OWL Reasoners: Locally or Globally?

Abstract. We propose a novel approach for performance prediction of OWL reasoners. It selects suitable, small ontology subsets, and then extrapolates rea-soner’s performance on them to the whole ontology. We investigate intercorrela-tion of ontology features using PCA. Finally, we discuss various error measures for performance prediction and compare our approach against an existing one using these measures.