Hi Elias,
Let me try a logical benefit in my opinion of defaulting to plain literals. Let's do a survey of *all* of the ontologies and RDF schemas we can possibly find and let's see how many rdf:XMLLiteral datatypes are found in them. If XMLLiterals win, then I think that should be the default, else plain literals should be. Is this logical?
I didn't ask for benefits...I specifically asked for a logical justification. My rationale is 'logical' because I have said 'given that we are processing an XHTML document, we are therefore dealing with XML literals BY DEFINITION'. What is the equivalent logical argument that says that not only is this view wrong, but also that we are actually dealing with plain literals? Don't get me wrong, I'll be quite happy if there is one...I've never minded being proved wrong! (I've had to get used to it over the years. :)) But so far, I'm not hearing an argument that keeps everything logical and consistent in both the XHTML and RDF camps. And counting triple datatypes is certainly not a logical argument, since the whole point I have been making is that RDFa is *not* driven by RDF, but by XHTML. Although even then, I have asked what difference it makes from the RDF side, that we use XML literals rather than plain ones.
You mention "almost without thinking about it". As much I'd like that to be the case, it's just not going to be "intuitive" to author RDFa w/o having a good grasp on graphs and RDF to write the least number of about attributes to fully exploit our subject resolution mechanisms. In my discussions with people about RDFa, I get a lot of push back in the form: it's crazy for me to think that the average person will know how to do this. In defense, I say the same about microformats, but that doesn't help us. In the end, we end up agreeing that most of the RDFa will be based on applications that have data in databases and output (X)HTML. Now if you think about it, how many systems allow mark-up in their fields? If a system does, then template should rdf:XMLLiteral as the content type, else it will most likely be a plain literal or some other datatyped literal.
RDFa is *totally* built on the belief that there will be 'RDFa authors', and that in fact these authors are crucial to the creation of a semantic web. So although you may believe that the only use of RDFa will be from databases, I'm afraid that horse bolted along time ago--that is not what RDFa is about. With respect, this illustrates to me why RDFa began life in the HTML Working Group and not an RDF group, since the HTML WG's primary concern was to get RDF from authors by causing them as little pain as possible, and not simply to come up with yet another way of serialising triples. In this case I can't believe we're suggesting that an author who needs to use mark-up in a string of text (such as a book title) must add both the datatype attribute *and* the RDF namespace to their code, otherwise the RDFa processor will obliterate their information. Regards, Mark -- Mark Birbeck, formsPlayer [EMAIL PROTECTED] | +44 (0) 20 7689 9232 http://www.formsPlayer.com | http://internet-apps.blogspot.com standards. innovation.
