- From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
- Date: Wed, 5 Dec 2001 12:49:22 -0000
- To: <w3c-rdfcore-wg@w3.org>
Attn all naughty bit suggests some part of URI space to be used for bNodes when serializing. Search for "NAUGHTY" Hi Dave, here is some serialization text possibly for sect 6 in your draft: [[[[ <p> There are two different approaches to serializing RDF. </p> <p> The basic approach uses the basic RDF syntax from RDF Model & Syntax([RDFMS]). In this: </p> <ul> <li> All blank nodes are assigned arbitrary URIs. </li> <li> Each resource is listed in turn as the subject of a top-level <code>rdf:description</code> element, using an <code>rdf:about</code> attribute. <br/> For each triple, with this resource as subject, an appropriate property element production is used, with either string content (possibly empty) or an <code>rdf:resource</code> attribute specifying the object of the triple. </li> </ul> <p> The basic serialization is recommended for applications in which the output RDF/XML is to be used only in further RDF processing. </p> <p> Where the intent is for the output RDF/XML file to be read by people, the basic serialization proves unsatisfactory. </p> <p>The basic serialization does not conform to more restricted sub-dialects of RDF, such as RSS [RSS] or CC/PP [CC/PP]. Hence, it is not appropriate for such applications, for which dialect specific serializers are needed. </p> <p> If more human readable output is required the following factors should be considered: </p> <ul> <li>There are many choices, with many RDF/XML documents corresponding to identical RDF graphs. Individual triples can be represented in numerous ways. High quality RDF serialization requires that these choices are considered by the serializing code. Some are more appropriate than others in an application dependent fashion. </li> <li>The triples in the graph need to be considered in an appropriate order. There are many choices of order, some being more appropriate than others in an application dependent fashion. </li> </ul> <!-- NAUGHTY bit --> <p> Serializations tend to be more readable if they do not attach URI labels to blank nodes. However, it is necessary, for some graphs, to do so, in order to refer to a blank node in two parts of the RDF/XML document produced. This is typically necessary when a blank node is the object of two triples. In such cases, RDF processors may use URIs beginning with "http://www.w3.org/2001/12/RDF/bnode/". </p> <!-- end NAUGHTY bit --> <p> It is not possible to use the RDF/XML serialization for serializing an RDF graph in which any triple has a property label which cannot be expressed as a qname. </p> <p>An approach to serializing RDF/XML using the full grammar in a top-down recursive descent fashion is discussed in [UNPARSING]. </p> ]]]] Additional references: [[[[ [CC/PP]Composite Capability/Preference Profiles (CC/PP): Structure and Vocabularies W3C Working Draft 15 March 2001, World Wide Web Consoritum http://www.w3.org/TR/CCPP-struct-vocab/. [RSS] RDF Site Summary (RSS) 1.0, Gabe Beged-Dov, Dan Brickley, Rael Dornfest, Ian Davis, Leigh Dodds, Jonathan Eisenzopf, David Galbraith, R.V. Guha, Ken MacLeod, Eric Miller, Aaron Swartz, Eric van der Vlist, 2000, http://purl.org/rss/1.0/spec [UNPARSING] Unparsing RDF/XML, Jeremy J. Carroll, HP Labs Technical Report HPL-2001-294 http://www.hpl.hp.com/techreports/2001/HPL-2001-294.html ]]]] (note URL for tech-report doesn't work yet - could omit last part) Jeremy
Received on Wednesday, 5 December 2001 07:52:31 UTC