1

Educational Modeling Languages,

From an Instructional Engineering Perspective

Gilbert Paquette

Center for Interuniversity Research on Telelearning Applications

CIRTA (LICEF), Télé-université, Université du Québec

Abstract

This chapter analyzes, from an instructional engineering (IE) viewpoint, the concept of an Educational Modeling Language (EML). The work on EMLs and its subsequent integration into the IMS Learning Design Specification is the most important initiative to date aiming to integrate instructional design (ID) preoccupations in the international standards movement. On the other hand, ID has evolved to what can be termed “IE”, integrating ID, software and cognitive engineering processes and principles. It has much in common with Educational Modeling Languages, sharing a software engineering and cognitive science approach. The main difference comes from the fact that IE is a methodology mostly concerned with the processes and principles that will produce good specifications of a learning system, in particular, EML specifications. Our goal here is to find out how an IE method could be adapted to and synchronized with a learning design specification standard like EML/IMS-LD and, conversely, how such a method could contribute to the evolution and use of a learning design standard.

The accelerating evolution of learning technologies has multiplied the number of decisions one must take to create a distributed learning system (DLS). While it is true that a majority of the first Web-based applications have been mostly ways to distribute information, more and more educators have become aware of the need to go beyond these simple uses of information and communication technologies. This context has created a much-needed interest in pedagogical methods and, more generally, the field of instructional design (ID).

In American literature, this discipline is known as "instructional design (ID)", "Instructional System Design (ISD)" or "Instructional Science" (Reigeluth, 1983; Merrill, 1994). In Europe, one of the pioneers in the field used the term "Scientific Pedagogy" [Montessori, 1958]. The origin of ID goes back to John Dewey, who, a century ago, claimed the development of an "interlinked science" between learning theories and educational practices (Dewey 1900). His demand was heard at the beginning of the 1960s, when we can speak of the beginning of a new discipline. In the 1970s and the 1980s, instructional theories have blossomed, but today, it seems necessary to renew the ID methodology to support the creation of distributed learning systems in order to operationalize the theoretical foundation.

Previously, the author has proposed a new approach to ID (Paquette, 2001a). This approach is founded on cognitive science and labeled as instructional engineering (IE), which is defined as a method that supports the planning, analysis, design and the delivery of a learning system, integrating the concepts, the processes and the principles of ID, software engineering, and cognitive engineering.

Software engineering, brings some interesting solutions to this goal. From a technical point of view, a DLS is an information system, a complex array of software tools, digitized documents and communication services. By adapting software engineering principles to ID, IE proposes well-defined processes and principles that help produce "deliveries", precisely describing the products of these processes. Moreover, multi-agent systems offer a good way to represent a DLS at delivery time as a set of agents, persons and computerized objects, interacting to help some of the agents to learn and others to facilitate learning.

Knowledge engineering is a methodology, developed in the field of expert systems and artificial intelligence over the last thirty years. It helps to identify and structure knowledge, to explain it, to represent it in a symbolic or graphic language, facilitating its subsequent use by persons and computer systems. Knowledge engineering has been applied in education to build intelligent tutoring systems (Wenger, 1987) and also support systems for designers (Merrill, 1994; Spector et al., 1993). There is now a renewal of interest in the integration of knowledge representation in the form of ontologies, as a basis for a new generation of the Web, the semantic Web (Berners-Lee et al, 2000). In an IE method, the knowledge engineering processes can help designers define content and objectives, instructional scenarios, instructional materials, as well as the delivery processes of a learning system.

A knowledge engineering approach is a response to the increased need for the reuse of knowledge resources and the interoperability of e-learning systems that has led to a vast movement towards international standards for learning objects (LOs). (Duval & Robson 2001).

The work on Educational Modeling Languages (Koper 2001), and the subsequent integration of a subset in the IMS Learning Design Specification, is the most important initiative to date to integrate ID into the standards movement. In particular, it describes a formal way to represent the structure of a Unit of Learning and the concept of a pedagogical method, specifying roles and activities that learners and support persons can play using LOs. Instructional Engineering, as defined above, and Educations Modeling Languages have much in common. They put the main emphasis on pedagogy and ID. They share a software engineering approach, EML being represented using the UML software modeling methodology.

In this chapter, an IE Method, MISA (Paquette 2001a) that shares similar goal with EML/IMS-LD will be presented. Our goal is to study how an IE method could be extended and synchronize with a learning design specification standard like the IMS-LD and conversely, how such a method like MISA could contribute to help designers using a learning design standard.

EML and the IMS Learning Design Specification

EML is described in Chapter XXX. As in the MISA method that we will present later in this chapter, the central concept of EML is that of a learning unit, an “Educational Modeling Language” being essentially a notation for learning units or units of study.

The IMS Learning Design Conceptual Model

The Educational Modeling Language developed at the Open University of the Netherlands (OUNL)has served as a basis for the IMS Learning Design (LD) Specification – Version 1.0 (IMS 2002). The approach in IMS-LD has been to define a complete core that is as simple as possible, with some extensions.

The Level A specification contains all the core vocabulary needed to support diversified pedagogical models. Level B adds properties and conditions enabling personalization. And Level C adds notification between actors involved in the learning unit. Figure 1 (reproduced from the IMS – Learning Design Specification document, version 1.0, p.10) presents a conceptual model of these three levels.


Figure 1 – The IMS Learning Design Conceptual Model

When activating a unit of learning, the method element is central. This unique element and its sub-elements control the behavior of the unit of learning as a whole, coordinating the activities of the players in their various roles and their use of resources. The Method, Plays, Acts and Role-parts are all nested within each other, as displayed in Figure 2, reproduced form the IMS – Learning Design Specification document, version 1.0, p.73.

There are three levels in a Method. At the first level, we find two elements, a list of plays and a complete-method object. The latter holds both the condition for completion of the unit-of-learning and optional actions to be taken when it is. The plays represent logically independent parts of the learning design as they are always run concurrently. They can be used to provide alternative scenarios for the same unit of study for different target populations or for different delivery models (e. g. classroom-based vs distance learning).


Figure 2. Structured Method in a Learning Design

An act brings together one or more role-parts to allow more than one role to perform at the same time or asynchronously in a certain time period. Therefore, role-parts within an act always run in parallel. Each role-part associates exactly one role with exactly one activity or environment. The same role can be associated with different activities in different role-parts and conversely. However the same role may only be referenced once in the same act.

Instructional Engineering with MISA

Figure 3 presents a high-level view of the main components of the MISA 4.0 IE Method and its relation to the Explor@ delivery system. In short, MISA supports designers, in formulating the design of an instructional system. This learning design can then be used to produce a runtime instructional system that can function within a delivery system such as Explor@ which is described in Chapter XXX (See also Paquette, 2001b), or another Learning Management System (LMS) or Learning Content Management System (LCMS). Figure 3 presents also the interrelations between the four main models created using MISA 4.0 as presented in Figure 1.

The MISA 4.0 Instructional Engineering Method

We have used a knowledge modeling approach to define the IE method itself, its concepts, its processes and its principles. This effort started in 1992 and has led to MISA 4.0 (Paquette 2001a, 2002a) and a Web-based support system for designers, ADISA (***) (Paquette et al 2001).

The root task: "To produce a model of a learning system" is distributed into six phases, each phase progressively developing four design models and their objects’ descriptions.

  • The Knowledge Model is a graphic representation of the content domain of a learning system. In this model, target and prerequisite competencies are associated to units of knowledge providing learning objectives to the Instructional Model.
  • The Instructional Model is essentially a network of learning units (LU) and events, to which knowledge and target competencies are associated. Each LU is described by a graphic learning scenario describing learning and support activities linked to resources. Resources holding content (called instruments) are associated with a subset of the knowledge model.
  • The Learning Material Models are optional. Each model groups instruments into a material (or LO), describing the media components, the source documents and some presentation principles and other specifications to build or aggregate LOs.
  • These learning materials and the other types of resources such as tools, communication links, services and locations, described in the Instructional Model, can be organized in one or more delivery model. Each delivery model is a multi-user workflow process where actors use or produce resources while adopting different roles.


Figure 3 – Interrelations between MISA 4.0 Design Specifications and Explor@

MISA 4.0 comprised 35 basic sub-tasks, each producing one design element. In a manner similar to software engineering methods, each model definition starts with a statement of orientation principles. In each of the four axes, one or more graphic models are built. Graphic modeling is the backbone of the method. It is done by a designer using the MOT (Paquette 2002b) knowledge representation technique and tool.

Most of the other tasks in MISA describe properties of objects in these models. For example, target and prerequisite competencies are properties of objects in the knowledge model. Learning activities and learning instrument design elements are properties of the objects in the learning events network or in the learning scenarios. MOT models and object properties all translate to XML. ADISA uses the corresponding XML files for data propagation and export, providing an XML binding for MISA design specifications.

3.2 The Instructional Model in MISA

As shown by the UML class diagram on Figure 4, at the highest level, an Instructional Model is a network of learning events.. “Learning Event” is a generic term to describe a module, a course, a training program, etc. A network oflearning events iscomposed of learning events, resources, links and rules. Composition (C) links build the hierarchy of learning events, while precedence (P) links describe prerequisites between them. Resources or products are related to learning events using input/product (I/P) links. Rules governing the use of learning events are connected to each other using a regulation (R) link. This is where, for example, it is possible to specify that there is a choice between three alternative learning events or units or that evaluation takes place in a certain way in a learning event.

A second step in the elaboration of the instructional model is to build a learning scenario for each learning unit. A learning unit is a learning event that is not decomposed into other learning events. It has learning objectives and prerequisites, provided by target and prerequisite competencies defined in the Knowledge Model.

Figure 4 – A class diagram of MISA 4.0 Instructional Model


A learning unit references exactly one Instructional Scenario grouping actors, resources, activities and links. Actors rule Activities (R link) that use or produce resources (I/P links). Activities can also be linked by precedence (P) links but no composition (C) links are allowed. As for a learning event, execution, evaluation, collaboration and adaptation rules can be R-linked to activities. The subset of the scenario grouping activities performed by a learner and their associated resources is called the Learning Scenario. The subset grouping activities performed by all the other actors (called facilitators) and their associated resources is called the Assistance Scenario.

Further subsets of the Assistance Scenario are possible for different types of facilitators; we could obtain a Trainer Scenario, a Content Expert Scenario, a Manager Scenario, etc. In the Virtual Learning Center Model on which the Explor@ Web-based learning delivery system is based, we have distinguished five main types of actors: learner, trainer, informer, manager and designer, but the system itself accepts other actor typologies (Paquette 2001b). In MISA, the possibility is there but it has not yet been implemented in the ADISA support system.

Five types of resources can appear in the learning events network and in any instructional scenario in a learning unit: instruments, tools, services, locations (where learning is carried out) and communication links (such as “broadband”, mail or face-to-face). These categories have been further decomposed in a typology expanding the IEEE LOM typology. In IEEE LOM, section 5.2 where the learning resource types are presented. The following are listed: exercise, simulation, questionnaire, diagram, figure, graph, index, slide, table, narrative text, exam, experiment, problem statement, self assessment and lecture. Of course the IMS-LD permits the extension of this typology, for example including learning units and methods (IEEE, 2002).

Instruments are the only resources that hold content. More precisely, they are associated to a sub-model of the Knowledge Model. We distinguish them from learning materials because they can, in general, be produced in different media formats. Usually, instruments are small pieces of information needed or produced in the activities of a learning scenario that will be grouped and implemented in a certain media format to create a learning material. Thus, learning materials, described in another model, are also associated with a knowledge sub-model. In particular, an evaluation material, such as a questionnaire, an exam or an essay, is also associated with a knowledge sub-model and the competencies linked to knowledge in that sub-model. These competencies are the basis on which evaluation takes place.

The author has shown elsewhere (Paquette 2001a) that it is possible to derive a learning scenario from a generic skill involved in a target competency, in the context of different educational models. For example, if a target competency states that learners should learn to diagnose equipment failures, the generic diagnostic process provides a workflow or task model composed of the individual diagnostic tasks with their inputs, products, and control principles. This approach is similar to the KADS software engineering methodology (Breuker et al, 1999). An instructional scenario is created when a pedagogical model is added to this basic flow of tasks. In an expository approach, an instructor will use the workflow model to present or demo segments of the diagnostic process. In a discovery approach, diagnostic problems in the field of equipment failure will be proposed to the learners, an instructor (or a learner) using the workflow model for assistance with the tasks.

Figure 5 presents an example of a MISA diagnostic learning scenario. In a MOT graphic model, ovals represent activities that are performed by actor roles (L for learner or F for facilitator). Rectangles represent resources labeled I for instruments, T for tools, S for services, C for communication or L for location. Unmarked resources are outcomes produced during an activity. Hexagons represent rules: X for execution, E for evaluation, C for collaboration and A for adaptation.

In the learning scenario subset (in white), learners rule six activities, starting with the analysis of an electronic system to diagnose for faults. A collaboration rule (C) states that they work in teams of 2. Execution rules (X) define iteration between activities until the complete system has been analyzed. Through these cycles, using LOs as inputs, each team produces intermediate outcomes and, finally, a list of defaults components. Using an assistance scenario (in grey), a facilitator distributes systems to teams, provides feedback using a forum and document transfer, providing evaluation services to learners, trainers and training managers.

Figure 5. An example of a MISA scenario


Comparing MISA and IMS-LD ontologies

As can be ascertained from the figures, there are many similarities between the main concepts of MISA and those of IMS-LD. Although some of the terminology differs, the IMS-LD subset of EML has approximately the same scope as the Instructional Model in MISA.

MISA as an Educational Modeling Language

In a study of Educational Modeling Languages, Rawlings et al. (2002, p. 10), defined EML as a “semantic information model and binding, describing the content and process within a ‘unit of learning’ from a pedagogical perspective in order to support reuse and interoperability.”

It is clear that the MISA Instructional Model constitutes an EML according to this definition. The set of MOT models grouping the Network of Learning Events, plus the Instructional Scenarios of each learning unit, and their associated knowledge models constitute a semantic information model that describes the content and process of any learning event from a pedagogical perspective. The translation of these MOT models into a set of XML files constitutes a semantic information binding that serves the same reusability and interoperability purposes as the XML files in the IMS-LD specification.