- From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
- Date: Tue, 23 Oct 2001 20:25:36 +0100
- To: w3c-rdfcore-wg@w3.org
Brian McBride wrote: > > The use case is to: > > o merge two graphs using different lexical representations of an int > We can use fiat to solve this use case: "1: All RDF datatypes MUST provide a canonical lexicalization s.t. a = b iff canonical-lex(a) = canonical-lex(b)" 2: A well-formed RDF/XML document MUST use the canonical lexicalization for each literal." Then an out-of-the-box M&S RDF will work. Somewhat ugly, but robust; well up-to rounding errors. I note that CHARMOD uses this approach to strings, where there is more than one equivalent lexicalization of certain (non-US English) strings, one (known as NFC) is the canonical form, and well-formed XML documents must use that one. Perhaps I am too uncritical of CHARMOD. Jeremy
Received on Tuesday, 23 October 2001 15:21:13 UTC