How Should We Think About Linked Data?

The semantic web tries to put the multidimensionality of human-type knowledge back into the web's flat mass of data. RDF tries to make this possible in a rather abstract way -- by means of triples, the subject-predicate-object group mapped roughly from natural languages (or rather some natural languages).

RDF triples link two things in some specific way. But even the formulation 'link things' suggests a kind of distinction -- thing-relation -- that cashes out in natural languages at times only rather problematically. And, of course, even just plain naming things is a slippery problem, somewhere at the foundation of logic and mathematics -- though a pragmatic 'triple'-type solution to this problem is something cyberneticists formulated long ago.

So in some deeper sense, the RDF triple model is kind of awesome. But, as Ian Davis points out, it can be confusing to grasp, and perhaps, for development purposes, doesn't need to be.

Ian's reformulation uses language familiar to developers, not logicians, defining a Resource Description as consisting of:

1. A subject resource identified by a URI
2. Zero or more relationships to other resources
3. Zero or more properties having literal values

A relationship comprises a pair of URIs: the first URI denotes the type of relationship, the second identifies the related resource.

A property comprises a pairing of a URI with a literal value. The URI denotes the type of property. The  literal value is a string with an optional datatype (denoted by a further URI) or language tag (as defined by RFC-3066)

Which doesn't require quite all the elaborate supporting materials offered by W3C.

As Ian points out, his description isn't really unique; he's simply trying to express a difficult idea more simply, by making it more technical and less abstract.

The comments are worth reading too (especially if you like SPARQL), so check it all out.