- From: <herman.ter.horst@philips.com>
- Date: Thu, 6 Feb 2003 20:03:19 +0100
- To: www-rdf-comments@w3.org
- Message-ID: <OFD0747BAA.1AA688F1-ONC1256CC5.00392C42-C1256CC5.0068D826@diamond.philips.com>
I agree with and would like to expand on Dan Connolly's remark about RDF graph equality. The RDF Concepts and Abstract Syntax document defines RDF graphs to be sets of triples. In this way, equality of RDF graphs is also defined. Namely, by one of the first axioms of set theory, two sets are equal if and only if they have the same elements. Therefore, the definition of RDF graph equality given in the Concepts document would introduce contradictions: two RDF graphs which do not have exactly the same triples may be equal and not equal at the same time. Equivalence would indeed be a perfect name for the notion that is defined. Note that equivalence thus defined is an equivalence relation on the class of all RDF graphs. I can understand the intent behind the word "equal" in the Last Call RDF Concepts draft. For many purposes, RDF graphs can be replaced by equivalent graphs. A central point that could be noted is as follows: the truth of any semantic statement involving RDF(S) (for example an interpretation I satisfies an RDF graph E, or a set of RDF graphs S entails an RDF graph E), is not changed when any RDF graph is replaced by an equivalent graph. In analogy with many similar situations in various mathematical theories, it could be said that "equivalent RDF graphs are identified", but then identical cannot be interpreted to be the same as equal. Herman ter Horst
Received on Thursday, 6 February 2003 14:05:16 UTC