Previous	SLIDE 16	Next

The Topic Maps Paradigm imposes certain "disclosure" requirements on ontologies. Ontologies must make certain commitments in order to participate in the mainstream of semantic integration that the Topic Maps paradigm seeks to establish. Ontologies have to be clear about, or provide a basis for being clear about, which subjects are to be reified, and which aren't.

Remember, there's no way to accumulate knowledge around a subject if it has no reifiers; if it has no reifiers -- no representations -- there's no place to accumulate it. Making sure that the subjects of reifiers can be recognized as being the same, when they are the same, is the price of the benefits that the paradigm provides.

One might think that a Topic Map Application should provide for the reification of all subjects, no matter what they are. Actually, however, since no topic map can be infinitely large, every topic map must involve some subjects that remain unreified. There are trade-offs here. The Topic Maps paradigm requires the trade-offs to be made, and these design decisions to be disclosed.

Ontological disclosures required

In your ontology (your "Topic Map Application Disclosure"), you must say how situations in which reifiers have the same subject will be detected.

More specifically: your ontology must define "Subject Identity Properties" capable of identifying all the subjects you wish to make reifiable.

Basic structural rules:

1. Every reifier must have at least one Subject Identity Property.
2. No reifier can have more than one Subject Identity Property that is defined by the same Topic Map Application.

Note: The latter rule establishes modularity for Topic Map Applications. A Topic Map Application is a module of merging logic. Topic Map Applications can include other Topic Map Applications as modules.