This page contains test data that we used in several of our papers.
Please refer to ontology authors' Web sites for licenses under which these ontologies are distributed.
GALEN is a medical terminology that has often been used for testing performance of description logic reasoners. The original version of the ontology is in the GRAIL language; however, a translation of the current version of GALEN into OWL can be found here. GALEN has traditionally been a very hard otnology for DL reasoners, mainly because it contains many cyclic general inclusion axioms with lots of existential quantifiers, which makes DL reasoners generate large models. Still, GALEN can be translated into Horn clauses, so no nondeterminism is needed for reasoning with it.
The current OWL version of GALEN is still too hard for all reasoners, including HermiT. In fact, some 10 years ago, the ontology was much smaller, but it was still too hard for DL reasoners; we call this version of the ontology GALEN-original. To obtain an ontology that can be processed in practice, GALEN-original was simplified by removing some 127 axioms; we call this version of the ontology GALEN-simplified. This simplified version of GALEN was used as basis for many benchmarks of DL reasoners. HermiT is currently the only DL theorem prover that can classify GALEN-original.
GALEN-module is a module extracted from the current OWL version of GALEN using the modularization algorithm from by Bernardo Cuenca Grau, Yevgeny Kazakov, Ian Horrocks, and Ulrike Sattler.
GALEN-module-no-inv is obtained from GALEN-module by removing all inverse property axioms.
BAMS (The Brain Architecture Management System) is an ontology about brain anatomy developed at the University of Southern California. Since HermiT still cannot deal with data properties, we have converted all data properties into object properties.
FMA-fragment is a fragment of the Foundational Model of Anatomy extracted by a student in the group of Alan Rector at the University of Manchester. Since HermiT still cannot deal with data properties, we have converted all data properties into object properties.
DOLCE is a foundational ontology developed at the Laboratory for Applied Onotlogy. Our version has been obtained from the original version by merging individual OWL files into one file and by converting all data properties into object properties.
Wine is an ontology that has often been used to demonstrate the capabilities of DL systems. Our version is obtained from the original version by removign one data proeprty.
NCI is an ontology developed by the National Cancer Institute. This version on our Web site is rather old: it is the result of an initial translation of NCI into OWL; we obtaied the ontology from mindswap. The ontology is very large, but also very simple. It can be translated into Horn clauses, so no nondeterminism is needed for reasoning with it.
NCI has evolved in the past years and the current version is more complex and uses more OWL constructs. It can be downloaded here.
Do not hesitate to contact any of the authors with questions about the prover. We are very happy to discuss reasoning algorithms and implementation techniques.