Standards » TMCL

In the latest draft we have something called topic-reifies-constraint, which is stronger than this. Using it, you could say that every topic of type X must reify a role of type Y:

X isa tmcl:topic-type;
must-reify(Y).

I assume this meets your use case, even if in a slightly different way?

Well, the rationale for having topic-type, association-type, etc is that this is the obvious way to define what, say, the topics "Person" and "Employed-by" are. You really must have these types to be able to make any sense of your ontology at all.

But "reifier-type"? Let's say I reify all my description occurrences to record more information about them. That means I make topics of type "Description", and I attach a topic-reifies-constraint to say what they must reify.

If we were to add a "reifier-type" then that would mean that these topics would have to have two topic types, right? "Description" and "reifier-type". Conceivably, one might make "Description" a subtype of "reifier-type", but it's not semantically correct (this would make all the description occurrences be reifier-types, too).

Another issue is that "reifier-type" doesn't actually add any information. You can infer that "Description" is a "reifier-type" from the topic-reifies-constraint. However, you could not do the inverse.

In short, I'm skeptical. However, we can discuss this at the next meeting and see what the general consensus is.

By the way, "Description" would have to be a topic type. So "reifier-type" must be a subtype of "topic type".

The Seattle meeting considered this, and rejected the idea of including such a constraint.