 |
Dock Allen is the project
lead for MITRE-Sponsored Research project, DiMeS: Distributed
Metadata Services. |
Making Sense of Metadata: The Benefits of XML
Dock Allen
October 2002
The Internet has a problem. And as MITRE's
Dock Allen sees it, we all have a problem because of it.
Say you've refined your Web search as much as is humanly possible
and still Google returns to you 3,000-plus URLs for your reading
pleasure. The problem is that the Web's cataloging format—called
metadata—is missing an information structure. For example,
if a card catalog is the difference between a library and a pile
of books on the floor, then the Web's current metadata structure
more closely resembles the pile of books. Dock Allen, a principal
software systems engineer, is on the front lines of making sense
of metadata—and has made considerable progress.
"The Department of Defense is experiencing the same kind
of exponential information growth as is the Internet," says
Allen, explaining the impetus driving her MITRE Sponsored Research
project, DiMeS (pronounced dimes, for Distributed Metadata Services).
"We're now using XML to put structure to content and also to
access other services." XML is a mark-up language, as is the
current Web standard HTML, but differs from HTML in that it focuses
on the data structure and content. HTML, on the other hand, focuses
solely on how a Web page looks.
The benefits of XML are obvious, as seen in this HTML/XML comparison:
(A) HTML keywords as metadata:
(B) XML brings order to the same keywords:
With XML, the pile of books begins to look more like a library.
Of course, once the catalog of cards—the XML metadata—is
ready for the library, it needs to be put where it can be easily
found and used. Allen, and her team of Paul Silvey and Laurie Hurwitz,
created a metadata-based "information broker," what to
XML developers is popularly known as Universal Description. This
is an information directory that is an easy-to-use mechanism that
helps clients dynamically find information and services.
Think Yellow Pages, as in a single source to fill all of the needs
for goods, services, and information—the information broker
lists it all. You make a request, and unlike Google's replies by
the thousands, your query returns a tightly combed list of exact
or near-exact responses. Or you can subscribe: instead of making
a query each time you need something, the information broker, based
on your pre-selected interests, routinely and automatically sends
you updates.
The Distributed Metadata Services team further expanded the concept
to distribute the metadata residing in these brokers by Communities
of Interest (COIs), which are groups of like individuals who share
similar products or services. "The Air Force Materiel Command
is a COI," explains Allen, "with its own special needs
as well as special language and terminology. It could have its own
customized and scalable information broker able to access other
information brokers across the entire network." Such a distributed
service simplifies the query and retrieval process, keeps COIs up
to date, and saves time, and avoids duplication of effort.
As the group's three-year project draws to a close, Allen is proud
of its accomplishments. "We've done all and more than we set
out to do. We've delivered the custom information brokers, a profiling
language with which to advertise broker services, and a software
development kit. In fact, all of this is out the door and in the
hands of industry, hopefully for final development and implementation
somewhere."
Allen will now migrate her expertise to other MITRE projects.
One in the offing, which she feels would be a great match, is a
Web Services project that directly bears on her XML metadata and
other software engineering expertise.
Allen came to MITRE about two years ago, after almost 30 years
as a software engineer. "I just love software," she says
with obvious joy, "and I just love the opportunity to do it
here at MITRE where I can do the right thing without the profit
push from industry always at my back. It is refreshing here."
|