Human Friendly Trix

Hi Jeremy, Patrick, www-rdf-interest,

Have been reading about TriX with interest. The primary client
interface at my work is browser javascript/xsl/xhtml, so being able
read and manipulate RDF using XML tools is a definite advantage of
trix over RDF/XML for me.

I do however agree with the 'xml in rdf' comments - there is a large
group of people who want their metadata to 'look' like xml, and
encouraging xml people to use RDF is of great importance to the
semantic web IMHO.

I've been experimenting with writing a simple xsl transform to convert
some human-readable xml-ish TriX into the native triples form.

An example of the xml is at:
http://www.phildawes.net/2004/01/trix/humanreadable.xml

The transform is at:
http://www.phildawes.net/2004/01/trix/humanreadable-trix.xsl

(N.B. your browser may attempt to transform this on the fly.
linux users can transform it using:
xsltproc http://www.phildawes.net/2004/01/trix/humanreadable.xml)

It (so far) supports qualified names (foo:bah), element-names as
rdf-properties and the rdfs:type in the outer element name. It doesn't
do nested or anonymous nodes yet.
E.g.
		<foaf:Person hr:qname="pd:phildawes">
			<foaf:nick>Phil</foaf:nick>
			<pd:commutesTo hr:qname="pd:city/London"/>
			<foaf:homepage uri="http://www.phildawes.net/"/>
		</foaf:Person>

Notes:

- I had to use the xslt namespace axis to convert qnames into
uris. This is unfortunate because not all xslt engines support this,
most particularly the one built into mozilla and firefox. Does anybody
have any tips for removing this dependancy?

- The namespace for TriX doesn't seem very persistence-friendly to me:
It's built using a domain name you don't control, and there's no date
identifier. Wouldn't something like http://purl.org/TriX/2004/02/ be
better?

- The xslt is pretty messy, due to bugs in my xslt transformer, and in
my understanding of xslt.

Comments?

Cheers,

Phil

Received on Monday, 16 February 2004 06:25:22 UTC