|
About CORDRA
Philip V. W. Dodds, J.D. Fletcher,
& Robert Wisher
The Advanced Distributed Learning (ADL) Initiative, sponsored by the Office of the
Secretary of Defense, is a collaborative effort involving government, industry and
academic participants. Its goal is to establish a new distributed learning environment
that enhances the accessibility of education, training, and performance aiding by
ensuring their interoperability and sharability.
ADL developed the Sharable Content Object Reference Model (SCORM) for creating and
deploying reusable and interoperable instructional content objects. Descriptions
of current versions of SCORM are available at the ADL site http://www.adlnet.org. SCORM integrates and harmonizes a
collection of specifications and standards adapted from global specification organizations,
accredited standards organizations, and consortia. It provides a comprehensive suite
of e-learning capabilities that will enable interoperability, accessibility, and
reusability of Web-based learning objects. SCORM is receiving global acceptance
as a standard for developing interoperable learning objects.
However, even with SCORM, we still need a means to identify learning objects uniquely,
locate them, and retrieve them on a scale that can support all users in a consistent,
persistent, reliable, and secure way. Overall we need a means to make learning objects
globally visible, while ensuring continued local control over access to them. These
are the concerns of the Content Object Repository Discovery and Resolution Architecture
(CORDRA).
The Problem Space
Over the past several years, developers of digital instructional content have begun
adopting an object-based approach based on relatively small “chunks” of instructional
material. The idea is for these content objects to be shared and reused in as many
different instructional situations as may suit the needs of individual learners.
Specifications such as SCORM suggest ways to “package” instructional objects by
adding information about their substance, structure, and rules for accessing them
so that information about the object can be discovered and searchable repositories
of objects built.
SCORM addresses several critical aspects of object design: how to identify, organize,
package, and move objects and how to track the learner’s progress and mastery, but
it is mostly silent on ways to make objects visible and ways to find them – specifically
ways to find relevant material that teaches a particular skill, or helps a user
solve a specific problem in a way suited to the individual learner.
SOME Assumptions and Requirements
The following assumptions and requirements describe working assumptions related
specifically to ADL’s distributed learning environments:
|
I.
|
Developers want their content to be accessed. We assume that instructional material
developers who make their content available in a repository intend the content be
found and used by others. “Publishing” content to a repository indicates that the
developer wants to make it discoverable and retrievable.
Requirement: A means to determine where relevant objects are available.
|
|
II.
|
Most users and developers are not skilled at either tagging content or expressing
detailed queries. Just because content has been packaged properly does not mean
that useful search information was also provided. Interpretations for when to tag
content and with what vocabulary vary widely.
Requirement: Guidance and very simple interfaces for tagging content.
|
|
III.
|
Searchers for objects have specific criteria in mind. It is assumed that people
who want instructional content know what they need. This might be expressed with
a simple description, key words, a specific skill or piece of knowledge, a relationship
to other processes, or the state of a learner’s profile.
Requirement: A means to relate context-based search criteria to descriptions of
specific content objects (e.g., mapping a skill definition to an object designed
to address that specific skill).
|
|
IV.
|
Searchers for objects want only what they need. Presumably searchers want to have
confidence that their selections are a precise match to their requirements. They
do not want every content object that “might” pertain. They want the ones that do
pertain.
Requirement: A means to ensure that discovered objects are relevant, accredited,
and authorized (among other qualities).
|
|
V.
|
A rigid information, service, and protocol model will limit its applicability. Experience
and studies have shown that distributed networks that rely on standardizing most
or all of their protocols, data models, and services are of limited applicability
– they do not “scale” well. This is because local systems, requirements, and practices
vary widely.
Requirement: An approach that is low cost and easy to implement and that allows
voluntary support and adoption.
|
|
VI.
|
The architecture must accept local policies and business rules, not prescribe them.
The approach must accommodate widely varying requirements and needs. Enforcing strict
procedural conformance will not be broadly supported.
Requirement: The means to institute and expose local business rules and policies
so they can be used or mapped to and from other systems.
|
|
VII.
|
We cannot foresee all the services or capabilities that will eventually be required.
We will not imagine all (or even most) of what might be needed or wanted for this
architecture.
Requirement: An architecture that allows new services and capabilities to be added
without changing the underlying structure.
|
These assumptions and requirements are flexible and intended to guide the evolution
and evaluation of specific solutions to serve contextually relevant discovery (precise
identification) and resolution (precise location) of appropriate learning objects.
Relating Context, Discovery, and Resolution
The process of obtaining instructional content that meets the needs of the learner
involves three processes – contextualization, discovery, and resolution – all of
which need to be integrated into a single coherent solution.
1. Context
We assume that a search for an object is driven by specific needs arising from a
particular learning need. In a simple case, a learner may want to find an object
that relates to a particular topic. In a more complicated case, the search for an
object must take account of the learner’s background knowledge, objectives, and
perhaps learning style combined with the subject matter, itself, learning environment,
instructional objectives, and appropriate instructional strategies.
For instance, imagine that automobile mechanics are learning to troubleshoot faults
in the steering mechanism of an unfamiliar car, but that they have been certified
to work on similar vehicles. A search for relevant content would want to know (a)
the make, model, and version of the vehicle with which the technicians are dealing,
(b) the skills needed to repair it, (c) the mechanics’ level of knowledge and skills,
and (d) the knowledge and skills needed for the specific vehicle at hand.
This scenario assumes that:
- A database exists somewhere with the exact configuration of
the steering mechanism of each vehicle.
- Someone has defined a skills taxonomy for the mechanism.
- A profile exists of the technician’s proficiency.
- Instructional content exists for the steering mechanism.
- Instructional strategies exist to prepare this specific technician
to troubleshoot this mechanism.
Assuming this information exists and is accessible, one can imagine the development
of a service or agent that can derive from these data the criteria needed to identify
instructional content that is contextually appropriate for the mechanic’s needs.
This context then provides the criteria required for discovery.
2. Discovery
The field of searching and cataloging is large and diverse. Library science tends
to focus on thoroughness of searches within specific collections, i.e., get all
the pertinent literature on a subject. Libraries often base their searches on catalogs
and indexes. Some collections are linked to one another by cross-referencing, shared
indexes, and common protocols.
Web search engines such as Google focus on indexing everything (get everything out
there that might pertain). Google-type searching “crawls” through thousands of pages
of World Wide Web text every day, indexing whatever it finds. The often noted result
is that Google can produce hundreds of “misses” for every relevant “hit.” Also,
there is a good deal of content it cannot find and index, such as non-text-based
material or repositories that are not connected to the Web. A means to find content
with more precision, accuracy, and confidence is needed.
Neither the structured approach to discovery used by the library community nor the
text-based approach used by Google will scale well to large numbers of applications
seeking the granular learning objects needed for instruction such as those being
considered by SCORM. Neither makes use of packaging and metadata specific to learning
(e.g., Learning Object Metadata at http://ltsc.ieee.org/wg12/index.html).
Ideally the discovery of learning objects should involve a process like the following:
- Develop search criteria from the local context
- Go to a master index of relevant repositories
- Go to one or more likely repositories
- Discover what objects are available in the repositories.
This approach suggests the idea of a registry of repositories — a place to go to
find out where the learning objects being sought can be found. If such a registry
existed, its index and metadata could be mined by a discovery service, i.e., software
could be developed and provided so that learners and developers could find precisely
the objects sought. Publishers of learning objects who wanted their content to be
found could voluntarily register their objects and provide information about their
content. Searches would be more precise because their scope would be narrowed to
intentionally published objects.
3. Resolution
Identifying an object and knowing where an object is are different things. Currently,
we rely on Universal Resource Locators (URLs) for location. These have been highly
successful tools, but they tie Internet resources to their current network locations
and to local file paths included with the URL. When the resource, or object, is
moved, the connection is severed. Uniform Resource Names (URNs) are persistent identifiers
for Internet resources not tied to specific network locations, but commonly used
browsers do not yet understand them.
The term “resolution” refers to more than just identifying the location of an object.
It also refers to linking an identifier to a variety of services and functions that
manage the use of the object. Authentication services can be used to determine who
can obtain access to objects. Other business rules can be enforced to protect intellectual
property rights. Life cycle management and maintenance policies can also be applied.
Resolution Using the Handle System
The Handle System (http://www.handle.net)
was developed by Robert Kahn and his team at the Corporation for National Research
Initiatives (CNRI) in the mid-1990s and can be used as Universal Resource Names,
among other things. The Handle System defines globally unique identifiers that can
be associated with information about an object and its location. The system defines
ways to build a registry of handles (unique names) and the services needed to resolve
a handle’s associated information (e.g., location). In the Handle System, the handle,
or unique name, is stored in a “handle server” along with a pointer to the object’s
location. The process of obtaining the location of the object is executed by a “resolution
service” that asks the handle server for the location information, among other things.
CNRI hosts a root “global handle server” that is a registry for other second-tier
handle servers. Several of these second-tier servers have been established as naming
authorities and offer packaged services, such as object location resolution, authentication,
application of business rules, and metadata storage and use. The packaging of these
services with the basic handle registry capability provides persistent storage and
retrieval of content objects.
A Proposed Architecture
As suggested by Figure 1, the principal components of a proposed learning object
repository discovery and resolution architecture (i.e., CORDRA) are context, discovery
leading to identification, and resolution leading to location, retrieval, and delivery.
Figure 1. Outline of the CORDRA Architecture for Learning Content
Contextualization, Discovery, and Resolution
1. Context: The Basis for Relating Specific Criteria to Searches for Objects
Context provides specific information about individual learner characteristics,
goals, and needs in a form that can be linked to meaningful search criteria. This
information may include domain maps, skill/competency taxonomies, knowledge ontologies,
learner profiles, key metadata, and the like. The ADL approach focuses on metadata,
learner profiles, and skill/competencies taxonomies. These are then used to provide
specific search criteria that can be used to find appropriate objects.
2. Discovery: A Process Using Context to Identify Repositories of Relevant Objects
and the Policies for Accessing Them
For communities with a large number of object repositories some means is needed
to make each repository “visible” to searchers. Discovery provides the means to
locate repositories and ways to search them for specific kinds of objects. ADL is
developing specific guidance for defining how repositories of education and training
objects should make their content searchable and the local policies and business
rules for doing so.
3. Resolution: A Process Using Context and Discovery to Identify Relevant Objects,
their Locations, and other Information Needed to Retrieve and Deliver Them
Once a mechanism exists to search repositories, the discovery process can reveal
objects’ identifiers. Resolution provided by the Handle System can then resolve
its location, metadata, and other necessary information as required. The object
can then be accessed in accord with local policies and business rules for doing
so.
Many approaches to repository and object management have emphasized one of the three
elements (context, discovery, resolution) over the others. All three must be considered
equally and in parallel.
The ADL plan is to remain on a relatively high level and concentrate on services,
data, and capabilities so that use can be negotiated rather than predefined. Different
education and training communities can then evolve somewhat independently of one
another but still be able to create access to objects they choose to share.
The CORDRA initiative was established in 2004 to address these issues and produce
specific implementation guidance to support new Department of Defense policy. This
policy requires that learning objects be registered in searchable repositories.
In accord with the original tasking for ADL, it is intended to stand as an example
for other communities to use as they see best in establishing their own policies
for sharing objects, instructional an otherwise.
The following references provide further information about CORDRA:
Dodds, P.V.W. (Ed.) SCORM 2004. http://www.adlnet.org.
Dodds, P. V. W., & Fletcher, J. D. (2004). Opportunities for new “smart” learning
environments enabled by next generation web capabilities. Journal of Education Multimedia
and Hypermedia, 13(4), 391-404.
Kahn, R.E., and Wilensky, R. (May, 1995). “A framework for distributed digital object
services,” [http://hdl.handle.net/4263537/5001].
Wisher, R. A., & Fletcher, J. D. (2004). The Case for Advanced Distributed eLearning,
Information & Security: An International Journal, 14, 17-25.
|
