RDF/XML Answers
This is a collection of answers by Stefan Ram on the topic of RDF as given in newsgroups.
Mixed Questions
The example some parts of the following message refer to was given by Jordan Katz. Because it was not quoted in the message, it is given here.
Consider the following example:
<?xml version="1.0"?>
<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/rdf-syntax-ns#"
xmlns="http://www.foo.net/jordan/index.html#"
xmlns:dc="http://purl.org/DC#">
<Homepage link="http://www.foo.net/jordan/index.html">
<dc:Title>Jordan Katz's webpage</dc:Title>
<dc:Creator>Jordan Katz</dc:Creator>
<dc:Date>2002-04-28</dc:Date>
</Homepage>
</rdf:RDF>
Now the posting follows.
Message-ID: <aabufo$9l0qj$1@fu-berlin.de>
Subject: Re: basic RDF/XML issue.
From: Stefan RamJordan Katz <katz@underlevel.net> writes:
> - are there substantial differences between the rdf:about and
> rdf:resource properties? From what I've read and from the
> examples I've seen they seem to be interchangable.
The attribute "rdf:about" is an attribute of the
element "rdf:Description", i.e., the subject.
The attribute "rdf:resource" marks an URI as
its value, in order not to mismatch the URI
for a literal text value.
rdf:about always needs a URI while rdf:resource
is used where a literal string might also be supplied.
> - how are properties defined? In other words, for the above
> example, I used the property "link." I know two ways of
> "defining" link (correct my use of terms here if I'm wrong.)
> One way would be to provide it with a name space but keeping it
> as a property. For example, if dc defined it I could use
> <Homepage dc:link...
> another way would be to use a tag within <Homepage, which would
> also require a namespace, like:
> <Homepage ...><dc:Link>http://www.foo.net/jordan/index.html">
Properties might be defined as XML-attributes
as well as as XML child elements. The meaning
in the model is not changed.
> - Finally, what "defines" the Homepage tag? What good is it if
> it's not given a namespace?
The element type name "Homepage" is implicitly typed. Using
rdf:type a class might be explicitly specified for an element.
However, instead of an explicit type, a namespace prefix
might be used to specify the type implicitly.
The xmlns=-namespace in your source applies to
all names without an explicit namespace, therefore
it specifies the class of "Homepage".
So, the type (determining the class) is:
http://www.foo.net/jordan/index.html#Homepage
Metadata and Triplets
Message-ID: <aaitfs$b0g0l$1@fu-berlin.de>
From: Stefan Ram
Subject: Re: Metadata questionsGustaf Liljegren <gustafl@algonet.se> writes:
>A search for "metadata" today results in large amounts of information about
>XML and especially RDF.
"Metadata" is just a buzzword for "data". (Actually, there
is some sense to it, but it is relative: One man's data is
another man's "metadata".) Moreover, "metadata" usually is
intended to be machine readable to some degree, while the
natural language text on a web page might be data about
some other web page, but it is not considered to be
"metadata", because it is in a natural language.
It is all about representing information in machine readable
form without a predetermined restriction on the nature of
the information.
For example, an array of floating point numbers allows one
to store only a sequence of numbers. An implementation of
a general entity-relationship-model on the other hand can
store all kind of information.
> I wonder if words like "triplet" and "resource" are
>universal terms common in _all_ metadata discussions?
Before I learned about RDF, I independently "invented"
triplets. One arrives naturally at triplets, when one tries
to dissect an entity-relationship-model (similar to a UML
static structure diagram) into elementary atoms of information,
all of the same size and structure, that can be handled more
easily than the whole thing.
A "resource" in RDF is similar to the notion of "entity" in
the entity-relationship-model: It can be everything with an
"identity". In a relational database, for example, the
primary key determines the "identity", while in RDF URIs
are used for that purpose. Because a URI can be made up for
everything one can speak about, "everything" can become
a resource.
> I also wonder if the
>object part of a triplet is the same as a "direct object" in English
>grammar?
It is vaguely similar, but not the same. An English sentence
may also have other parts of information (like the indirect
object or specifications of location, time, quantity and so
on). A triple always has three parts. An english sentence
might have to be dissected into several triplets.
Metadata
In a reply to the message above, Mark Preston claims Stefan Ram 's explanation of metadata to be wrong. His words are not archived here, because he has set the X-No-Archive-Flag in his posting. So the following is just the answer of Stefan Ram to Mark Preston with the quoted part from Mark Preston 's posting removed.
Message-ID: <aajbip$b5nf8$1@fu-berlin.de>
From: Stefan Ram
Subject: Re: Metadata questions"Metadata is machine understandable information for the web."
http://www.w3.org/Metadata/