HCLSIG BioRDF Subgroup/Tasks/URI Best Practices/Recommendations/ShorterSputnikDraft/DraftTalk

"Assert owl:sameAs when not already asserted for fully equivalent terms. (x owl:sameAs y means that x may be replaced by y and v.v. in a graph without changing meaningfulness or truth value.)"

SameAs is *equality*, thus applied only to "individuals" (if one wants to stay in OWL DL). EquivalentClass and EquivalentProperty are biconditionals.

I think equality and equivalence are very heavyweight mechanisms for vocabulary alignment. They pretty much allow you to *drop* one of the terms (i.e., rewrite all your axioms to use just one). They don't allow for variant definitions, for example. Now the rest of the document seems to be against using the same term with variant definitions, but it also doesn't say what to do when you *find* a term with multiple definitions.

Since it considers rdfs:comments to be adequate definitions (and the primary required definition) what do we do if the comment and the supplied axioms disagree? Am I, a term consumer, allowed to reuse the term that has only a comment with my own formalization of the comment? Are these two definitions or one?

Not *every* term *can* have (at least usefully) a definition (we need a primitive vocabulary). Now, perhaps we're always grounding out in natural language (which seems sensible), but a bit more detail would be happy.

Similarly, I don't know if there's a precise notion of definition at work here (e.g., necessary and sufficient conditions which is at least clear, but with known general problems).