Problems with OWL Syntax

In this paper we discuss three problems with OWL syntax that repeatedly surface in practice. The first problem is that OWL does not allow for explicit declarations—assertions that a certain class, property, or an individual exists in an ontology. This aspect of the OWL standard was often misinterpreted, which caused design errors in OWL APIs; moreover, the lack of declarations makes devising an intuitive structural consistency check for OWL ontologies difficult. The second problem is that OWL Abstract Syntax and OWL RDF syntax rely on the separation between object and data property names for disambiguation. We show that this prevents an unambiguous interpretation of certain syntactically well-formed OWL ontologies; furthermore, it makes implementing OWL parsers unnecessarily difficult. The third problem is that OWL Abstract Syntax cannot be translated into OWL RDF syntax without loss of information. We present possible solutions to these three problems, which, if adopted in OWL 1.1, would lead to a cleaner standard and would significantly simplify the implementation of OWL APIs.