- markup (76)
- xml (11)
- xslt (23)
- pipelines (8)
- atom (9)
- overlapping markup (6)
- schema (11)
- creole (5)
- dtll (1)
- xforms (1)
- xpath (1)
- xquery (2)
- coding (2)
- datagovuk (1)
- genealogy (4)
- hardware (1)
- linked data (15)
- named graphs (1)
- opendata (1)
- psi (3)
- skos (1)
- sparql (4)
- Talis (7)
- unicode (1)
- uri (4)
- versioning (1)
- visualisation (6)
- web (77)
- google (4)
- html5 (5)
- jQuery (2)
- rdf (45)
- ontologies (2)
- rdfa (8)
- rdfQuery (5)
- rest (6)
- wikis (1)
- work (3)
- legislation (2)
- xmlsummerschool09 (2)
- life (28)
- children (5)
- equality (6)
- gtd (1)
- environment (4)
- gadgets (5)
- software (3)
- xlinq (2)
- conferences (11)
- ukgc09 (1)
- xtech (9)
- xtech2008 (3)
- blog (8)
- drupal (3)
Re: What You Can't Do with HTML5 Microdata
Uh, OK. Given that a significant proportion of the people who work with the semantic web and who want to use metadata embedded within web pages are currently using RDF, I would have thought you’d use a “pave the cowpaths” approach. But I know your mind is made up so I’m not going to try to change it.
Regarding datatyping, you might consider using at least those datatypes that have an obvious mapping in Javascript, since this will allow people to compare and manipulate values (such as numbers) within their scripts without converting them explicitly. And with dates and times in particular, it would seem to break the principle of least surprise to have the type indicated within the markup but not reflected within the DOM.
Regarding XML Literals, the word ‘XML’ may send you running and screaming for the hills, but they’re actually about capturing structure rather than using a particular technology. They are a useful feature for document content. For example, descriptions of events may well run to several paragraphs, or include emphasised text or ruby markup, and it would be good in these cases for the value of the property to reflect that structure.
As far as validating the values goes: it’s not necessary for an RDF processor to validate the triples that it holds in order to process them (in the same way as it’s not necessary for an XML parser or an XSLT processor to validate a document in order to build a model of it and transform it). Knowing the datatypes of values becomes important when you’re processing RDF in exactly the same way as it would when processing the values in Javascript: in running comparisons and performing calculations, in RDF’s case within SPARQL or when using a library such as rdfQuery.
Only OWL supports the definition of datatypes for properties, so you would have to use an OWL validator to actually validate the triples against a vocabulary. Validators are harder to find than you might hope; they usually exist within triplestores rather than as standalone applications.