The time take to index an ontology on loading can be increased by the need to resize indexes as they are constructed.
An extra data segment containing index cardinalities could be used to record the right sizes for data structures, to avoid dynamic resizing.
(Alternatively, indexing can be delayed in OWLAPI until after an ontology is loaded, so the sizes can be rebuilt).
Binary OWL could also provide its own OWLOntology implementation, with indexes persisted along with contents.
The time take to index an ontology on loading can be increased by the need to resize indexes as they are constructed.
An extra data segment containing index cardinalities could be used to record the right sizes for data structures, to avoid dynamic resizing.
(Alternatively, indexing can be delayed in OWLAPI until after an ontology is loaded, so the sizes can be rebuilt).
Binary OWL could also provide its own OWLOntology implementation, with indexes persisted along with contents.