- From: Garret Wilson <garret@globalmentor.com>
- Date: Thu, 09 Aug 2007 18:09:44 -0700
- To: Semantic Web <semantic-web@w3.org>
Everyone, I promise I'm not trying to dig up the "replace literals with normal resources" discussion, although I'm still strongly for that proposal. I have an honest question about plain literal semantics, however. At http://www.w3.org/TR/2004/REC-rdf-mt-20040210/#urisandlit I read: "Plain literals are considered to denote themselves, so have a fixed meaning." Is this a typo? It's not the most natural sentence---does it mean, "and thus always have a fixed meaning", or "and and so have no fixed meaning?" Next question: how do plain literals differ semantically from typed literals with a datatype URI of xsd:string? Yes, I know that they are "different" in the URI abstract syntax, but that's begging the question---what does that *mean*. They "denote themselves", I'm told---and what "themselves" are are strings of characters. What I'm asking is: in real-life practical applications, where would it ever be useful to talk about a literal absent its type information? That is, where would I ever want to choose a plain literal over a typed literal of datatype xsd:string? If were were creating RDF from scratch today, and we had already created typed literals, would we say, "Oops, we're missing something. We need something called a "plain literal" that has no associated type information?" I just don't see it... Of course, when I'm creating RDF/XML it's much easier for me to create plain literals. I'm just as lazy as the next person. But that's just a syntactical issue---would anybody miss anything if future RDF parsers were to interpret <eg:property>text</eg:property> as a typed literal with datatype xsd:string? (Sure, existing queries on plain literals would need to be changed, but that's not the question.) I'm not just trying to stir up more trouble, or to rail against the current RDF. This is a legitimate question, because if there is no reason (besides my laziness in typing RDF/XML) to choose plain literals over typed literals of datatype xsd:string, shouldn't I use the latter exclusively in all of the new ontologies I create? Rather than utter the sacrilegious, "should we deprecate plain literals in RDF?" let me ask it in a less forceful way: wouldn't be a good idea for someone to consider it best practice to exclusively use typed literals with datatype xsd:string over plain literals when creating new ontologies? Thanks in advance for any comments to clear this up. Best, Garret
Received on Friday, 10 August 2007 01:10:06 UTC