Metadata

standard bibliographic tags) can be extended
by more specific classes that add or replace
specific properties such as entries specific to
an art collection. New kid on the block      In May, the World Wide Web Consortium
issued the Resource Description Framework
as an official recommendation precisely to
address these issues. RDF is a metadata coding
scheme, expressed in extensible markup lan-
guage (XML). It enables the analysis of struc-
tured documents—and to some extent knowl-
edge representation—although it does not
explicitly specify a mechanism for reasoning.
According to the recommendation’s authors,
RDF corresponds to a simple frame system.
Figure 3 recasts the earlier example to show
what RDF can do that META tags cannot.      Figure 3 demonstrates how RDF can
describe any document’s properties access-
ed by an URL address. The example lists
properties drawn from four different vocab-
ularies, or classes, listed at the beginning of
the document under the XML name space
tag (xmlns): • RDF, the standard vocabulary for de-
scribing Web resources, which uses the
Description, Description about, and
value tags. • DC (Dublin Core), a set of standard
bibliographic properties defined by the
library community. • CRYPT, a hypothetical set of crypto-
graphic properties used for document
authentication and security audits. • ART, a hypothetical set of tags used by
art collectors.      With name spaces, you can use
the same property name, such as
Author, in two or more different
class contexts, without confusion.
Conceptually, the name spaces de-
fine classes of properties, and you
can interpret each tag within the
RDF:Description context as a Sub-
ject (About URL), Predicate (Class
Name:Property Name), and Object
(Property Value) statement.      Although not shown here, RDF
tags let you specify the properties of
individual sections of Web docu-
ments or entire collections of URLs,
information on how to parse • property values (for example, dates and measurement units), • nonatomic property values such as
nested structures, • lists of alternative values, and • ordered and nonordered sequences. You can also express predicates about predi-
cates. Reified statements such as “Person B
states that Person A is the Creator of URL
U” are expressed by adding an “attributed-
To” property to the RDF description: <RDF:Description about=”U”> <DC:Creator>Person A</DC:Creator> <ATTRIB:attributedTo>PersonB </ATTRIB:attributedTo> </RDF:Description>      RDF is a syntax specification
that enables a rich semantic de-
scription of Web-page content.
However, it is beyond RDF’s scope
to provide directions on how to
infer the meaning of and reasons
for such descriptions. For example,
what happens when two RDF de-
scriptions about the same URL
provide conflicting information? Is
it possible to infer other properties
beyond the ones found in an RDF
description? RDF does not directly Figure 3. RDF description for the same content described in Figure 2. 
<?xml version="1.0"?>

<RDF xmlns:RDF = "http://www.w3.org/1999/02/22-rdf-syntax-ns#"       xmlns:DC = "http://purl.org/DC"
      xmlns:ART = "http://auctions.org/artwork-tags"
      xmlns:CRYPT = "http://crypt.org/signatures">

<RDF:Description about = "http://www.mycompany.com/personal/journal.html"> 
<DC:Title>Some Document Title</DC:Title>
<DC:Author>Someone</DC:Author>
<DC:Rights>My Company</DC:Rights>
<DC:Date>1999-07-01</DC:Date>
<DC:Subject>Work, Life, Art, Dreams</DC:Subject>
<DC:Description> A day in my life, with a description of a painting by          Jackson Pollock for sale.
</DC:Description>

<CRYPT:Digest>          <RDF:value>f96b697d7cb7938d525a2f31aaf161d0</RDF:value>
         <CRYPT:Encoding>MD5</CRYPT:Encoding>
</CRYPT:Digest>

<ART:Painting>          <ART:Author>Jackson Pollock"</ART:Author>
         <ART:Type>Oil On Canvas"</ART:Type>
         <ART:Size>100 cm x 300 cm</ART:Size>
         <ART:Price>$1,000,000"</ART:Price>
         <ART:Date>1968</ART:Date>
</ART:Painting> </RDF:Description>
</RDF> JULY/AUGUST 1999 21