Pierre-Antoine CHAMPIN wrote: > > I think this raises a question about the semantic relation between Literal and Resources. It would be nice to let the literal BE the resource content. The XML representation of RDF would use literals to insert data for the resource. The resource could be implied, or explicitly stated. That would make the RDF implementation cleaner. There would be a natural way to store the content of any object in a triple: as a object with a URI. The implicit URI for literals would be derermined the same way as other implicit URIs. The content of URLs cold be stored in the object. There is no need for a difference between a literal (like a name of a author) or a URI (pointing to a name of the author ). Consider this: There is no way to know how to parse the literal, exept by metadata. The literal could contain the binary data of an image. Just like other resources, the literal is the subject of a triple specifying the type of the literal. The type can be any object that is a subclass to the object Literal. The URI for "image/gif" could be one of those literal sub classes. There is no way to know how to dereference a resource, exept by metadata. The resource could contain the binary data of an image. The similarity to literals goes on. (compare with the paragraph above.) The only difference is one of convinience for the XML representation of RDF. It is convenient to give a way to embed actual data in the web of metadata. The distinction between resources and literals is an effect of this syntax. But the distinction becomes blured on a higher level. The RDF Schema specification is a part of this, as it handles literals as being instances of the Literal class - that relation in itself beeing represented in RDF triples. * On one hand, you have to know how to parse a literal * On the other hand, you have to know how to retrieve a URI I suggest: - that ALL URI's representing retrievable data, will be considerd literals. - that each mime type will be a sub class to Literal The retrieval and parsing of data has to be supported by the implementation. The above suggestion support the decision on how to parse the data. For the XML representation of RDF, there is a clearly defined way to retrieve the embedded data. But for other URIs, there should be a way to know how to retrieve the data, as discussed here: http://paranormal.o.se/perl/proj/rdf/schema_editor/letters/resources.txt