96 Computer
S T A N D A R D S
I magine a world where everyoneexperiences “just right” learning:just the right content, at just theright time, in just the right con-text, in just the right medium.
Imagine this massively customized
and personalized “me-learning” for
each of the 6.4 billion unique individ-
uals in the world, 24/7.
Our vision for the future of learning
is one of mass customization and, ulti-
mately, personalization. While it
might be difficult to imagine such a
world, it is perhaps even more diffi-
cult to realize that a fundamental
enabler of such uniqueness and per-
sonalization is standards.
“Surely, you must be joking” is an
expected response, but this vision is a
fundamental driver for a collection of
standards within the IEEE 1484 Learn-
ing Technology Standards Committee.
The LTSC’s scope is to develop stan-
dards for learning technology—not
for learning itself. The first LTSC stan-
dard, Learning Object Metadata, was
completed in 2002 and several more
standards have followed since then.
THE LOM STANDARD
In the LOM context, learning objects
are small, reusable content units rele-
vant for learning. Metadata describes
the learning objects’ attributes, making
it easier to find, evaluate, acquire,
assemble, and deploy relevant content
(E. Duval et al., “Metadata Principles
and Practicalities,” D-lib Magazine,
Apr. 2002, pp. 1-16).
As Figure 1 shows, the core of the
multipart LOM 1484.12 standard is
a conceptual structure for a metadata
instance based on a hierarchy of nine
categories of metadata elements.
For every metadata element, the
1484.12.1 LOM data model defines
the number of values allowed, indi-
cates whether the order of values is
significant, and identifies the set of
allowed values and the data type. In
this context, data types are structures
based on other standards such as ISO
8601 or ISO/IEC 10646 or specifica-
tions such as vCard.
LOM instances can include exten-
sions to the base schema. The explicit
goal is to evolve the standard when
experience in practice demonstrates
a widespread adoption of such exten-
sions.
In practice, communities such as the
US Advanced Distributed Learning
initiative or the European ARIADNE
Foundation translate their needs and
requirements into application profiles
that make metadata elements manda-
tory, add extensions, or restrict value
spaces.
By adopting LOM, these communi-
ties can exchange metadata at a basic
level of interoperability. Through the
application profile, they can meet their
internal requirements at a higher level,
for example, by requiring that an
object has a title in all the official
European languages.
LOM accommodates multilin-
guality for both the learning object
content and the metadata. For ex-
ample, every string value can be
repeated for multilanguage metadata
instances.
In 2005, IEEE 1484.12.3, the stan-
dard for the XML binding of LOM,
was finalized as a set of XML schemas
that a particular community can con-
figure to satisfy its needs.
THE KEY TO SUCCESS
Accredited standards serve a pur-
pose and make a profound difference.
History has shown that the takeoff
point for many innovations includes
the adoption of common standards.
Examples include railway track gauge,
telephone dial tones, videotape for-
mats, e-mail protocols, the Internet,
and the World Wide Web. Without the
adoption of common standards, the
market stalls. Consider the historic
battle between VHS and Beta that hin-
dered the eventual explosion of the
video industry, or more current exam-
ples such as the lack of common stan-
Standardized
Uniqueness:
Oxymoron or Vision
of the Future?
Erik Duval
Katholieke Universiteit Leuven
Wayne Hodgins
Autodesk Inc.
The IEEE’s 1484 LTSC provides
a collection of standards for
customizing and personalizing
learning.

March 2006 97
35, www.slais.ubc.ca/PEOPLE/faculty/
tennis-p/dcpapers2004/Paper_15.pdf).
Similar to early Web authoring tools
that showed the details of HTML
tags, many early LOM tools were
basically electronic forms that closely
followed the LOM standard. Only
recently have tools become available
that automate the metadata authoring
process, using personal profiles of
metadata and contextual information,
as well as mining the learning object
for metadata.
Similarly, search applications increas-
ingly rely on a simple Google-like inter-
face to process queries over a federated
worldwide network of repositories. For
end users, boundaries—whether phys-
ical, technical, or organizational—be-
tween the repositories are irrelevant,
and they consider all of the material to
be part of the global collection of learn-
ing content.
Taking this approach further, we can
now evolve from searching to finding
and presenting appropriate content
when needed, without requiring an
dards for DVDs (Blue-Ray versus HD
DVD) and instant messaging.
Today more than ever, to be true
enablers, standards must be flexible
and adaptive so that they can be
increasingly useful as circumstances
inevitably change—often unexpect-
edly so.
In the work on LOM and other
LTSC standards, a key attribute to
success has emerged: flexibility—keep
the standards small and modular.
LOM provides an example of such
flexibility, offering modularity in two
significant ways. Every LOM meta-
data element is optional. Reaching
agreement on this purposeful decision
took several years. LOM effectively
standardizes how to structure meta-
data about learning objects, not which
metadata elements to include. Many
successful standards incorporate this
same flexibility attribute, including
HTML, where most tags are also
optional. Yet these are true standards
in that once the authors make a deci-
sion about what to express, the stan-
dard provides a clear, consistent, and
unambiguous way of doing so.
LOM ensures maximum flexibility
and longevity because it is a multipart
standard. To ensure maximum modu-
larity, interoperability, choice, and
adaptability, we separated the standard
for the individual metadata elements—
LOM’s data model (1484.12.1)—from
the technical binding or expression of
these metadata elements with XML
(1484.12.3). We believe this modular-
ity also accounts for why LOM has
thus far exceeded all expectations of
success as it has been globally and uni-
versally adopted, adapted, and imple-
mented in many unexpected ways and
applications.
WHERE TO FROM HERE?
Achieving LOM’s full potential re-
quires meeting the challenge of devel-
oping and promoting the adoption of
tools that make the metadata com-
pletely transparent to end users
(“Making Metadata Go Away—
Hiding Everything But the Benefits,”
Proc. Int’l Conf. Dublin Core and
Metadata Applications, 2004, pp. 29-
end user to know to request it.
Context plays a paramount role here.
For example, an RSS feed can include
relevant content for a professor’s aca-
demic course or personalized profes-
sional training can be part of an
employee’s desktop environment.
A similar evolution with related
metadata formats like Dublin Core
(from the more conventional library
domain) or MPEG (from the audio-
visual domain) mirrors the progress
and deployment of metadata that the
LOM standard supports. The evolu-
tion toward a Semantic Web offers
tremendous potential for merging all
metadata in a seamlessly integrated
information environment. This is cru-
cial for learning applications as mate-
rial that was not originally purposely
developed for learning can yield much
relevant content.
A s metadata becomes more per-vasive, automatically generated,and transparently processed, we
Figure 1. Learning object metadata structure.This structure is based on a hierarchy of
nine categories of metadata elements.
cost
copyright and other information
description
interactive type
learning resource type
interactivity level
semantic density
intended end user level
context
typical age range
difficulty
typical learning time
description
language
educational
rights
kind
resource
relationidentifier
description
catalog
entry
entity
data
description
annotation
purpose
taxon path
description
keyword
classification
source
taxon
id
entry
version
status
contribute
identifier
title
language
description
keyword
coverage
structure
aggregation level
general
life cycle
identifier
contribute
metadata schema
meta-metadata
role
entity
data
catalog
entry
format
size
location
requirement
installation remarks
other platform requirements
duration
technical or composite
type
name
minimum version
maximum version
Learning
object
metadata
catalog
entry
role
entity
data

98 Computer
S T A N D A R D S
will find that, rather than identifying
what, how, and who, searching will be
more like the recent IEEE slogan:
“Click less, find more.” In the future,
each situation’s context will define the
right content, and the information will
increasingly find the user.
Although transforming this vision
into reality requires substantial addi-
tional work, remarkable progress has
been made during the past decade,
including the wide-scale adoption of
IEEE standards such as LOM.
As change tends to occur exponen-
tially, it is possible to imagine a world
where personalized “just right” learn-
ing for all based on flexible metadata
standards is the norm. Standardized
uniqueness is clearly not an oxymoron
but rather the secret to successfully
realization of the vision of personal-
ized learning for each unique individ-
ual, in unique situations, everyday,
everywhere.
For those of us working on IEEE
standards, ensuring that we are con-
sistently improving our efforts be-
comes imperative as we strive to make
this vision of learning a reality. ■
Erik Duval, professor at Katholieke Uni-
versiteit Leuven, is a technical editor of
IEEE LTSC 1484.12. Contact him at
erik.duval@cs.kuleuven.ac.be.
Wayne Hodgins, director of worldwide
learning strategies at Autodesk Inc., is
chair of IEEE LTSC 1484.12 LOM. Con-
tact him at wayne.hodgins@autodesk.
com.
Editor: John Harauz, Jonic Systems
Engineering, Inc., Willowdale,
Ont., Canada; j.harauz@ieee.org
To submit a manuscript for peer review,
see Computer’s author guidelines:
www.computer.org/computer/author.htm
Computer
magazine
looks ahead
to future
technologies
• Computer, the flagship publication of the IEEE Computer Society,
publishes peer-reviewed technical content that covers all aspects of
computer science, computer engineering, technology, and
applications.
• Articles selected for publication in Computer are edited to enhance
readability for the nearly 100,000 computing professionals who
receive this monthly magazine.
• Readers depend on Computer to provide current, unbiased,
thoroughly researched information on the newest directions in
computing technology.
Welcomes Your Contribution