Index of /2001/04rs22

	Name	Last modified	Size	Description
	Parent Directory		-
	Makefile	2001-10-10 19:55	1.6K
	README.html	2001-05-30 23:26	5.9K
	aboutEachC1-noAE.rdf	2001-05-18 23:06	909
	aboutEachC1-noCol.rdf	2001-05-18 23:06	558
	aboutEachC1.rdf	2001-05-18 20:52	399
	aboutEachImp.rdf	2001-05-21 15:03	933
	aboutEachRules.n3	2001-05-21 15:03	802
	aboutEachSpecEx.rdf	2001-05-18 20:52	414
	aboutEachSugar.xsl	2001-05-18 20:52	3.3K
	colSugar.xsl	2001-05-18 21:07	2.2K
	confusedPrefixes.rdf	2001-05-24 14:55	475
	dc-filter-maker.n3	2001-05-21 15:03	393
	dc-filter-top.n3	2001-05-21 15:03	100
	dc-filter.n3	2001-05-21 15:03	3.8K
	dces.rdf	2001-05-18 23:06	9.6K
	fixAttrs.xsl	2001-05-24 14:55	1.4K
	fixedAttrs.rdf	2001-05-24 14:55	589
	litSugar.xsl	2001-10-10 19:35	1.5K
	rdf-mathlit-noLit.nt	2001-10-10 19:55	12K
	rdf-mathlit-noLit.rdf	2001-10-10 19:55	7.0K
	rdf-mathlit.rdf	2001-10-10 19:55	573
	rdf0.rdf	2001-04-17 05:48	87
	rdf0.xsd	2001-04-17 05:48	1.1K
	rdf0e.rdf	2001-04-17 05:48	112
	rdf1.rdf	2001-04-17 05:48	293
	rdf1.xsd	2001-04-17 05:48	2.5K
	spec.xsl	2001-05-25 19:38	80K

RDF Syntax: An XML Schema/XSLT Approach

Using the online XSV service, we can see the results of checking that rdf1.rdf conforms to rdf1.xsd. Checking rdf0e.xsd shows that it does not conform.

see also: earlier (Aug 2000) approach, among the semantic web advanced development materials.

Lowering the Syntactic Sugar in RDF Documents

The XML Schema approach above only covers the simplest parts of RDF syntax. But much of the rest of the syntax of RDF 1.0 doesn't add any expressive capability. Just to prove the point (and to reduce the burden on some of my tools that read RDF) I've implemented (the start of) an XSLT transformation that eliminates RDF collection syntactic sugar: colSugar.xsl. The Makefile shows how to run the simple test case that it passes:

aboutEachC1-noCol.rdf: aboutEachC1.rdf colSugar.xsl
	$(XSLT) aboutEachC1.rdf colSugar.xsl $@

Using our online XSLT service, you can try it out on any RDF document available via HTTP:

<rdf:RDF>
  <rdf:Description about="http://www.w3.org/Home/Lassila"
                   s:Creator="Ora Lassila" />
</rdf:RDF>

fixedAttrs.rdf: fixAttrs.xsl confusedPrefixes.rdf
	$(XSLT) confusedPrefixes.rdf fixAttrs.xsl $@
	$(TIDY) -i -m -xml $@

An interpretation of aboutEach

Consider aboutEachC1.rdf:

  <rdf:Bag rdf:ID="books">
    <rdf:li rdf:resource="#book1" />

    <rdf:li rdf:resource="#book2" />

    <rdf:li rdf:resource="#book3" />
  </rdf:Bag>

  <rdf:Description rdf:aboutEach="#books">
    <dc:rights>all mine!</dc:rights>
  </rdf:Description>

In the RDF 1.0 spec, we see...

No explicit graph representation of distributive referents is defined.

3.3. Distributive Referents: Statements about Members of a Container

But the use of URIs to point from a bag of properties to the collection of things that have those properties suggests that aboutEach could work across files. I suggest we consider rdf:aboutEach to be two things: (1) a bit of syntactic sugar, and (2) a rule that applies to each member of a collection.

The sugar represents a collection of properties and objects as a description element with an aboutEach attribute; it's implemented (at least in part) by aboutEachSugar.xsl. To see this sugar unwound, see the next stanza of the makefile:

aboutEachC1-noAE.rdf: aboutEachC1-noCol.rdf aboutEachSugar.xsl
	$(XSLT) aboutEachC1-noCol.rdf aboutEachSugar.xsl $@
	$(TIDY) -i -m -xml $@

The rule for what this graph representation of aboutEach implies is:

<> log:forAll v:nx, v:ny, v:rdesc, v:col, v:p, v:o, v:elt.

{ v:nx a rdfs:ContainerMembershipProperty.
  v:ny a rdfs:ContainerMembershipProperty.

  v:rdesc v:nx [ rdf:predicate v:p; rdf:object v:o ].

  v:rdesc rdf:aboutEach v:col.

  v:col v:ny v:elt.
}
  log:implies { v:elt v:p v:o }.

@@write-up interrupted here