I haven’t (yet!) given a lot of thought to the business of integrating ontologies. However, I fully expect that my experiences in extracting RDF from XML (described in detail elsewhere) allow me to anticipate challenges with redundancies and inconsistencies.
In the RDF-extraction-from-XML case:
- Redundancies are multiple instances of the same subject-predicate-object tuple
- Inconsistencies are multiple subject-predicate-object tuples for a single subject
In the context of ontology integration:
- Redundancies are semantically equivalent instances of a single object in two or more of the ontologies being integrated
- Inconsistencies are semantically distinct instances of a single object in two or more of the ontologies being integrated
In the RDF-extraction-from-XML case, it soon becomes evident that an iterative approach is required to address redundancies and inconsistencies. This will also be the case for ontology integration.
Automating such a process will also present a very interesting challenge!