W3C home > Mailing lists > Public > www-rdf-interest@w3.org > October 2000


From: Matola,Tod <matola@oclc.org>
Date: Mon, 9 Oct 2000 16:30:30 -0400
Message-ID: <72B89459DD2BD211B5CD0000F840094E020E7D90@oa3-server.dev.oclc.org>
To: "'www-rdf-interest@w3.org'" <www-rdf-interest@w3.org>
Don't know if you have seen this, but it is good press...:-)
Cheers Tod...


 <http://www.seyboldreports.com/> SeyboldReports.com	 

The Seybold Report on Internet Publishing

Vol. 5, No. 1 - September 2000

 <http://content.seyboldreport.com/SRIP/about.html> Home

 <http://content.seyboldreport.com/service/Subscriptions.html> Subscribe

 <http://content.seyboldreport.com/Search/> Search

 <http://content.seyboldreport.com/SiteMap/> Site Map

 <http://content.seyboldreport.com/about/ContactUs.html> Contact Us

 <http://content.seyboldreport.com/back.html> Back Issues

The Latest Word

Classifying the Web: Glimmer of Hope For an Indexed Web

Are you ready for the "Resource Topic Description Map Framework"?

The likelihood of a workable standard for Web metadata rose substantially
last month when a scheduled "shootout" between the W3C's Resource
Description Framework (RDF) and ISO's Topic Map (TM) standard at the GCA's
Extreme Markup conference in Montreal turned instead into a lovefest. A
standard for Web metadata would be very helpful in making sense of the
tangle the Web has become. It is possible to search for content now, but the
results are haphazard. As Tim Bray, co-editor of the original XML spec and
now with antarcti.ca, likes to say, "The Web should work more like a library
and less like a heap of books on the floor." A widely implemented standard
for metadata would make it possible to publish your site map in a way that
is universally readable and available to aggregators, search engines and
portals without customization. 

Background. The first major effort to create a markup standard for
classification was the publication in February 1999, of the Resource
Description Framework (RDF) by the W3C. RDF has its proponents, and some
have done extensive work with it. For example, the Online Computer Library
Center ( www.OCLC.org <http://www.oclc.org/> ) is using RDF in an impressive
array of applications that include work with the Dewey Decimal System and
the Library of Congress, as well as other specialized vocabularies. 

Overall, though, we still have yet to see a major surge in adoption of RDF.
The recent delivery of tools for handling RDF site summaries (
http://purl.org/rss/1.0/ <http://purl.org/rss/1.0/> ) and an open source RDF
metadata server that has been added to the Mozilla project (
www.xml.com/pub/2000/08/09/rdfdb/index.html) may signal the beginning of an
uptake in interest. 

The premise behind these standards: the Web should work more like a library
and less like a heap of books on the floor. 

RDF is part of the W3C Metadata Activity whose goal is 
"to produce a language for the exchange of machine-understandable
descriptions of resources on the Web." The initial recommendation was a
Model and Syntax Specification, published by the W3C in February 1999. It
uses directed graph notation to describe resources and can be represented in
graph notation or in XML. (The spec and plenty of associated material are on
the W3C site ( www.w3c.org <http://www.w3c.org/> .) 


Topic Maps (officially ISO 13250:2000) are relative newcomers, although the
expertise on the ISO committee spans decades of work in hypertext and markup
systems. Topic maps are the outgrowth of a committee's effort to apply
HyTime to descriptions of large knowledgebases. The spec has some strong
adherents and some early tooling in Europe and the U.S., but no major
implementations. Topic Maps are syntax independent; to date, they've been
implemented in SGML and companion ISO standards. An active working group
hosted by the GCA IDEAlliance is preparing XML Topic Maps (XTM) for the
purpose of putting Topic Maps on the Web. 

After the conference, the XML Topic Map group met and divided into three
subgroups to produce a conceptual model, usage cases and the XML syntax.
According to Eric Freese, director of product services at ISOGEN,
harmonization with RDF is a firm requirement for all three groups. Ideally,
according to Freese, the two specs will merge, but for the time being, they
are targeting harmonization, such that an XSLT script could be written to
convert between RDF, XML and XTM. 

Matching them up. In his summary compare/contrast, Eric Miller, senior
research scientist at OCLC, compared RDF resources to TM topics; RDF schemas
to TM templates; RDF properties to TM facets and association roles and RDF
URIs to TM topic identifiers. Both specifications have a typing system,
entity relationships and similar goals, but there are differences. Topic
Maps are not Web-specific and will not have a Web-specific semantic until
the XTM effort is completed. Unlike RDF, TMs link to actual occurrences of
concepts within a resource. 

One thing these standards do share is a deceptive Alice in Wonderland
quality about them: after five minutes, you think you've "got it," and after
an hour, you wonder where you are, how you got there, and if anything will
ever make simple sense anymore. With Topic Maps, topics and associations are
easy to conceive, but layering themes, rules, templates and facets quickly
go beyond the intuitive. RDF's stark abstraction spares us the terminology
overload, but its syntax is considered obtuse even by adherents. 

Among the possibilities held out by Miller are doing a Topic Map layer on
top of RDF.One large win for Topic Maps would be the use of open-source RDF
tools. Might the XTM effort design a better RDF serialization syntax than
RDF? If the joint work goes so far as identifying the common inference rules
and query language, it would be a big win for users. 

Combining RDF and topic maps would be a tremendous win for the user

Reaction. User sentiment at the conference was unequivocal: if having one
standard is better than none, having two is worse. The response to possible
convergence of these was strong, positive and influential, given the
luminaries in the audience. C. Michael Sperberg-McQueen, co-editor of the
original XML Recommendation and chair of the W3C Schema working group, spoke
from the floor and made the point that users have influence if they make it
known that they will not tolerate dueling specifications. Sperberg-McQueen
compared the relationship between RDF and topic maps to physics and
chemistry, where RDF is concerned with the atomic level of representing
reality, and topic maps are concerned with the molecular compositions. In
his closing keynote he came back to this theme and, while a W3C partisan by
association, he warned that "there is no `better living through physics'
slogan," meaning that the conceptual purity of RDF is not more important
than the utilitarian approach of topic maps. 

Jon Bosak of Sun Microsystems, who chaired the original XML work within W3C
and has since taken up the work of ebXML, pointed out that the two
specifications are aimed at different semantic levels, but, if they were
combined, it could be a tremendous win for the user community. 

The XML topic map group plans to wrap up its activities by the December XML
2000 conference. Let's hope this is more than a casual summer fling and that
the separate parties can do what it takes to make this relationship last. 

Liora Alschuler 

top> of Page
Home <http://content.seyboldreport.com/SRIP/about.html>  | Subscribe
<http://content.seyboldreport.com/service/Subscriptions.html>  | Search
<http://content.seyboldreport.com/Search/index.html>  | Site
<http://content.seyboldreport.com/SiteMap/index.html> Map | Contact
<http://content.seyboldreport.com/about/ContactUs.html> Us | Back
<http://content.seyboldreport.com/back.html> Issues
Received on Monday, 9 October 2000 16:30:33 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:07:32 UTC