Basics of the Resource Description Framework

Basics of the Resource Description Framework

We often hear that XML tags add meaning to documents. And this is true, but it’s meaning that only humans can intuit. For example, given the XML markup

<person name=”Jane”> <sells product=”books”/> </person>

a human might use the intuitive tag names and attribute name/value pairs to infer that “Jane sells books” (among other things that people may do), but this is a leap of faith that a machine cannot make. True, the content models in a DTD or schema can enforce that sells nests within person, but it does not tell us anything about why the nesting takes place.

Similarly, given the preceding markup, an XSLT style sheet could transform it into the following string:

“Jane  sells  books”

However, to the computer, this is just a string like many other strings. In the end, the computer has no power to do anything with the string other than display it. In particular, it can’t make any logical connection between the string “Jane sells this book” and the string “My human wants to buy this book” and make the purchase. Such strings have meaning only to humans.

What we need is to go beyond the notion of a content model to a “meaning model,” which is what the RDF data model provides. Listing 23.1 gives us the markup for the RDF statement that has the meaning “Jane sells books,” where “Jane” is in RDF’s sub-ject position, “sells” is in the predicate position, and “books” is in the object position. In this way, we use markup technology to tell the computer where the meaning is. Note that “[Jane]” and “[books]” are URIs, and sells is a name in the “[my]” namespace; you’ll see why later.

LISTING 23.1    A Simple Statement in RDF

<rdf:Description about=”[Jane]” xmlns:my=”[my]”> <my:sells rdf:resource=”[books]”/>

The idea here is just as simple as when a teacher draws a subject/verb/object diagram on the chalkboard in grade school, as shown in Figure 23.1. If you want your sentence to mean anything, you have to put the words in the right order—and you and your listener have to know what the order is.

What the teacher’s syntax production does for students, in chalk, the RDF data model does for machines, in bits and bytes. Now let’s look a bit more deeply into the three words that comprise RDF: resource, description, and framework.

Why resource? A resource is anything that has identity (Jane, for example). How do resources get identity? Through being identified by Uniform Resource Identifiers (URIs). Therefore, that which has identity is a resource, and that which is a resource has identity!

If you look back to Listing 23.1, you will see that both the subject and the object of the RDF statement are resources, because they are identified by URIs.

Not all resources (for example, Jane) are retrievable across a network. A URI can iden-tify a Web page, of course, but also a printed book, a government agency, a human being, or an abstract concept.

Why description? A description is really just a container; it is a bucket for one or more statements. This idea, too, is pretty much straight from the grade school dictionary. If I ask Jane’s friend (or maybe even her computer) for a description of Jane, I expect to get back a number of statements, one of which might be that “Jane sells books.”

Why a framework? Natural languages such as English and Esperanto permit speakers to generate infinite numbers of sentences, to invent new words, and to give new meanings to old words, all based on a reasonably small set of rules. These rules comprise the framework of the language. In the same way, RDF sets rules that will enable humans and machines to make and understand infinite numbers of statements whose subjects and objects are resources.