Learning objects and process interoperability.

CURRENT SITUATION

There has been considerable emphasis on the availability and reuse of learning content in recent years. Since 2000, the ADL initiative has refined the recommendations contained in the SCORM documents through progressive stages represented in the SCORM 1.0, 1.1, 1.2 and 1.3 documents. Fundamental to SCORM is the notion of the Shareable Content Object (SCO) and the Learning Object Metadata (LOM). There is an expectation that the Learning Experience can be designed using a set of Learning Objects or SCOs drawn from repositories of learning materials.

The use of eLearning technology has been hampered by the lack of appropriate tools to support the many processes of learning. There is also an incomplete understanding of the knowledge acquisition processes in learning and as a consequence there is an inadequate representational framework to support tool design. The standards incorporated in SCORM include the definition and structure of learning material expressed as SCOs, the metadata used to describe SCOs, the interface between the SCO and the Learning Management System as represented by a Data Model and an Application Programme Interface (API), the Packaging of SCOs, and the sequencing of SCOs to the learner to achieve a specified learning objectives. From the user standpoint, deciding on how to design SCOs in the first place is not a consideration for SCORM. Design of SCOs is an instructional design task which is carried out prior to the consideration of how they are sequenced to the learner. There are LCMS products to assist with the design of learning content but few attempt to provide support beyond the management of media assets in the formation of learning objects.

Equally, providing assistance to the user in selecting appropriately defined SCOs from a repository is not explicitly part of the process. SCORM assumes the SCOs exist, they are described by IEEE LOM metadata and they can be packaged according to the IMS Packaging Specification and the are accessed by the learner in an order defined by the IMS Simple Sequencing Specification. If all this is in place the packaged SCOs can be run on any SCORM compliant LMS thus achieving the essential content interoperability requirement. However, much of the process of putting together the learning experience remains a manual one or, if the user has an LCMS some proprietary functionality is available to "design" and organise the SCOs.

Obstacles to Progress

While SCORM has made significant progress in providing a framework for interoperability of learning content among LMS brands it does not address the issues of creating and using rich learning experiences. There are deficiencies in SCORM in that it does not have a way of describing the processes associated with learning other than Simple Sequencing. Processes relate to all aspects of learning from the design of the experience to the participation in the processes of learning, which must involve discourse among individuals.

Whatever aspect of learning we consider it is possible to distinguish user processes associated with all classes of users. Teachers go through processes to design learning experiences for a given set of learners. Learners go through a set of processes when interacting with the teacher and the learning materials as part of a learning experience. Teachers use scholarship to draw together materials on a set of topics, interpret the materials, and add their own interpretations in the context of the learning experience they are designing. In the actual design of learning materials the teacher may interact with the instructional designer, the media technologist, and the programmer in preparing a learning experience. A full analysis of these processes would reveal the nature of the support required for each stakeholder.

In addition, we require a more effective way to describe learning materials if we are to successfully combine learning components in a conceptually coherent way. Acquiring knowledge requires a framework into which concepts can be placed relative to each other and which are linked with relationships involving descriptions of order, type, part, and so forth. Knowledge derives not simply from the syntactic structure of the relationship between learning objects but must involve some form of higher level semantic description of the relative position and usage of the learning objects in the structure of knowledge of the domain. The current expectation is that learning materials can be obtained from local and remote repositories according to the specification of the learning experience being designed. But to achieve this the process of specifying the types of learning that will, at the same time, enable coherent matches to be made to existing content held in repositories will also require metadata that goes far beyond that available today as specified in the IEEE LOM.

What Needs to Be Done

To realize the full potential of ICT support of learning and knowledge acquisition we need to have two to achieve two objectives. First, we require a richer description of learning content and second, we require a framework in which to structure the description of supporting processes for all stakeholders. Achieving the first will help us to relate the learning content to the knowledge structure of the domain of discourse. Achieving the second will provide the opportunity for vendors to create new tools and services to stimulate innovation in all aspects of the learning lifecycle.

Semantic description of learning resources. The concept of ontology has previously been used to represent domain knowledge. In its basic form, ontology defines the common words, concepts, or meanings used to describe a domain of knowledge. Ontology is usually agreed among the members of a community as an acceptable way of describing the conceptual understanding of the domain. Taxonomy is part of ontology in that it represents the classifying description of the entities in the domain. Taxonomy therefore provides part of the "knowledge" of the domain by classifying the entities. Taxonomies carry with them a weak form of semantics of the domain in that they imply a structural relationship between the entities classified in the hierarchy.

Ontology contains far richer information on the meaning of a domain. An ontology not only contains information about the structure of the domain but also relationships between entities, properties of entities, processes involving the entities, and constraints and rules about how the entities relate and exist in the domain. Building ontology of a domain is about capturing in a representation the structure, meaning, rules, constraints, and relationship between the entities in the domain. Capturing and representing all this in a processable form is the reason for the excitement currently surrounding the developments associated with the Semantic Web. We believe the same requirement exists for the semantic description of learning content. We need to relate the description of learning content to the structure of the knowledge of the domain to which it applies.

There are now ontology building and management tools emerging in the market. With these tools it will be possible to express the semantic meaning of data, documents, processes, and systems in a machine processable form. In which case it will be possible to have software processes reason about entities in the domain and make meaningful decisions about them. For information on the Web this means the difference between a search resulting in tens of thousands of items, most of which are not relevant to the reader, and a result providing four or five highly relevant items. But this is only the most basic use of an ontology when considering the future of web services the availability of ontology's will provide opportunities to reason about the domain and the resources it contains. This opens up the whole area of ontology driven applications, which will move us to the next level of smart working with Internet technologies.

Description of processes and interoperability. An emerging approach to the design and structuring of computing systems today is the Service Oriented Architecture (SOA). The basis of the SOA is web services. Web services are built on the technologies associated with the Extensible Markup Language (XML) according to the standard technology stack. The various layers in the technology stack can be considered to represent levels of abstraction built upon the Core Layers. The three basic layers of interest to this discussion are respectively, the Infrastructure layer, the Services Layer, and the Application Layer. For learning technology systems there are several early approaches based on this architecture notably, the IMS Abstract Framework, the MIT Open Knowledge Initiative, and the work from the Learning Systems Architecture lab at CMU. In addition several of the major eLIG companies have adopted the SOA approach.

The SOA is specifically concerned with the definitions of processes, their coordination, and the information that flows between them. In terms of the technology stack mentioned earlier, the technologies WSDL and WSFL are part of the Services Layer, and these are used to define the abstract representations of processes and their interactions. For learning technology systems our starting point is the analysis of processes to support all aspects of the learning lifecycle. Some of these processes will be candidates for automated support tools and services while others will remain as manual tasks carried out by the stakeholder. The significant advantage of the SOA approach means that the behaviours of and interfaces to processes can be defined abstractly, independent of language and platform. This fact provides the opportunity for vendor implementations to innovate in the design of tools and services and yet remain interoperable with tools and services from others.

A full analysis of the processes to support all stages in the learning lifecycle will reveal where ICT can best be used to support each stakeholder. The ISO SC36 LTA Group identified more than 120 stakeholders in their approach to define a framework for learning technology architectures. An analysis is therefore likely to reveal a multitude of processes some of which would be usefully automated and others remain as manual tasks. A useful outcome of such an analysis would contribute to the definition of process behaviours and interfaces. Only when these definitions are available, in the abstract form provided by WSDL technologies, will it be possible to have a full stimulation of the market opportunities for innovation in tools design and implementation.

Analysis of Processes

To provide an example of the main principles of process interoperability we can consider a simplified situation involving several actors associated with the early stages of the development of a learning experience. Our basic premise is that we should first consider the set of lower level processes that form the activities associated with the basic workflows involved in designing the learning experience. Some of these processes will be human centered, for others it will be possible to map the required activities onto software specifications that can later be designed as products to be provided by tool and service vendors. For those processes or sub-processes that can be considered as candidates for tools and services we believe it would be an advantage to find abstract descriptions of the processes to allow the interfaces and the behaviours to be specified independently of the implementation of the tool. With such an arrangement it will be possible for multiple vendors to each provide their own interpretation and implementation of the tool or service. Vendors will be in a position to provide useful support tools and services and yet competitiveness in the markets will be maintained. A significant advantage to the customer will be the choice of multiple vendors implementations of the tools to meet the support of the given process stages. We maintain there will be advantages to users in being in a position to have interoperable tools in this sense.

Investigation Objectives

The overall objectives are fourfold.

1. to remove the ambiguity and fragmentation that exists in the current work on e-Learning Standards for Content Description Interoperability;

2. to provide a framework for the development of new tools and services using the SOA;

3. to develop a rich semantic description of learning content within a knowledge framework supported by ontology; and

4. to promote the design of new tools and services within selected knowledge domains to demonstrate innovation in supporting all stakeholder processes.

Phased Programme

A survey of current eLearning projects would provide an assessment of the current work relating to these objectives and avoid unnecessary duplication. There are other process oriented approaches to learning design incorporated in the EML framework (Koper, 2001) and the IMS Learning Design Specification.

New tools and services should be designed within the proposed service framework and these could usefully be focused on selected domains. Involving users of the tools and services in the definition of processes would be beneficial. A consultation process would reveal useful common approaches that would benefit the community if new tools and services resulted.

The work on eLearning standards and interoperability is in need of being extended to match the requirements of the new service enabled architectures being considered by several major vendors. Process interfaces and behaviours need to be defined so that vendors can demonstrate interoperability of supporting tools and services.

Ontology looks like a promising approach to provide the knowledge framework within which to develop the new semantic descriptions of learning content and learning sequences.

Pilot projects in chosen knowledge domains will provide the opportunity to demonstrate the effectiveness of the new approach especially if vendors are asked to provide tools for abstractly defined processes.

References

Koper, R. (2001). Modelling units of study from a pedagogical perspective--the pedagogical meta-model behind EML. Report, June 2001, Retrieved September 29, 2005 from http://eml.ou.nl/introduction/articles.htm

KEITH D. BAKER

University of Reading--Reading, UK

[e-mail]