Printer Friendly

Towards evolutional authoring support systems.

The ultimate aim of this research is to specify and implement a general authoring framework for content and knowledge engineering for Intelligent Educational Systems (IES). In this context we attempt to develop an authoring tool supporting this framework that is powerful in its functionality, generic in its support of instructional strategies and user-friendly in its interaction with the author. The framework that we offer is an ontology-based authoring environment, since we see the use of ontologies as effective means to have structured content and knowledge engineering. The key contribution of our research is the meta-level Authoring Task Ontology (ATO) specifying authoring tasks, goals and activities, and knowledge about the process of engineering IES. It enables us to enable the specification of "evolutional" authoring support system, as a meta-authoring tool that defines and controls the configuration and tuning process of an authoring tool for a specific authoring process. The role of ATO is to help the IES authoring system to "evolve" by defining such a meta-tool, which "knows" the system's ontological structure. In our approach it works analogously to an authoring tool when it generates a concrete learning support system. The ATO also provides a shared vocabulary between all system components, and allows for better interoperability in a modularised architecture. In this way we have the benefit to monitor and assess the authoring process, and to prevent and solve inconsistencies and conflicting situations. Through ontological engineering, a computational formalization of the intelligent systems authoring, we also give our scientific contribution to the in-depth understanding of what the authoring process is.



Intelligent educational systems (IES) have proven to be useful and have obtained a significant place in the field of courseware. As authoring of IES is the subject of this article we take a closer look at their authoring aspects. In the analysis of various educational systems (Brusilovsky, 1995; Mizoguchi, Ikeda, & Sinitsa, 1997; Murray, 1999), such as SmartTrainer (Chen, Hayashi, Kin, Ikeda, & Mizoguchi, 1998), AIMS (Aroyo & Dicheva, 2001), InterBook (Brusilovsky & Eklund, 1998), and AHA! (De Bra & Calvi, 1998), we observe that providing user-oriented instruction shows different aspects that are relevant to an author, for example, subject domain modeling, user modeling, instructional and learning strategies, sequencing and structuring the learning material and its maintenance.

The authoring process implies the construction of applications in this IES format. Building an IES requires a lot of work and often it is done from scratch. The current way of designing an IES offers little space for the reusing or sharing of content, knowledge, and functional components. A central problem in maintaining the popularity of IES and benefiting from wide use of IES in practice, is the fact that the current approaches for building IES are rather inflexible and not efficient. The high and dynamic user demands in many aspects of software production are influencing research in the field of intelligent educational software as well (Major & Ainsworth, 1997). The ultimate problems are related to keeping up with the constant requirements for flexibility and adaptability of content and for reusability and sharing of learning objects and structures (Devedzic, Jerinic, & Radovic, 2000) Another problem in current IES research is that assessment of the existing systems is difficult as there is no common reference architecture, nor standardized approaches.

Many researchers in the field of educational systems have been focusing lately in the field of authoring systems and their improvement (Kiyama, Ishiuchi, Ikeda, Tsujimoto, & Fukuhara, 1997; Redfield, 1997; Vassileva, 1995). It is now well known and proven that the popularity of IES and other knowledge-based learning support systems strongly depends on the usability and easy maintenance of their authoring systems (Brusilovsky, 1995). Although the research field has already identified the main requirements for the authoring of IES, still only very application-dependant ones exist and these do not focus on reusability neither of the development efforts, nor in terms of the applicability in different domains (Murray, 1999). As the application domains are multiple and serve various needs, the benefit of a common reference architecture would be significant.

On the positive side, we see that currently a considerable amount of the research on knowledge-based and intelligent systems focuses on concepts and ontologies (Devedzic, 2001; Mizogcuhi & Bourdeau, 2000; Vasilakos, Devedzic, Kinshuk, & Pedrycz, 2004) and focuses on knowledge sharing and reusability (Chen, Hayashi, Kin, Ikeda, & Mizoguchi, 1998; Ikeda, Seta, & Mizoguchi, 1997). In general, an ontology is used to define the basic terms and relations in the domain. Next to this it also provides axioms as rules and constraints for manipulating and managing the terms and their relations within this common domain vocabulary. Ontologies allow the definition of an infrastructure for integrating intelligent systems at the knowledge level, independent of particular implementations, thus enabling knowledge sharing (Breuker & Bredeweg, 1999). Together with various reasoning modules and common knowledge representation techniques, ontologies can be used as the basis for development of libraries of shareable and reusable knowledge modules (which take the form of software components) (Aroyo & Dicheva, 2002; Aroyo, Dicheva, & Cristea, 2002). As a consequence this research that focused on ontologies, offers tools and technologies for the reusing and sharing of knowledge and hence helps intelligent educational systems move towards semantics-aware environments (Aroyo & Dicheva, 2004).

So, when we observe that many of those knowledge-based instructional systems share common architecture features, and we have the recent results in the use of ontologies, then we conclude that it has become possible to specify a reference architecture for concept-based intelligent educational systems. In this way we can allow for a more structured and common approach in the authoring process, as well as supporting the automation of the authoring activities.


Our reference framework provides the functionality to bridge the gap between author and authoring system in two directions (in line with evolutions in the related research areas): managing the increased intelligence and the need for (conceptual) user-friendliness.

First, we look at the increased intelligence. Authoring of IES is a process with an exponentially growing complexity and it requires many different types of knowledge and considering various constraints, requirements, and educational strategies (Nkambou, Gauthier, & Frasson, 1996). Going towards a process of (semi)-automated IES authoring we need to have explicit representations of strategic knowledge (rules, requirements, constraints) to be able to reason within different contexts and situations within the authoring process. Authoring systems provide a common framework (in terms of functional syntax and semantics) to be instantiated with the needs and requirements of different types of educational systems.

Second, we consider the conceptual distance between user and system. From the analysis of the current state of the art of the educational systems research (Mizoguchi & Bourdeau, 2000; Redfield, 1997) it appears that there is a deep conceptual gap between authoring systems and the authors. The authoring tools are neither intelligent nor user-friendly. Special-purpose systems provide extensive authoring guidance, but the disadvantage here is that changing such systems is not easy, and the knowledge and content can hardly be reused for other educational purposes (Murray, 1999). All this leads to the requirement that the authoring tool should offer the user the possibility to configure or tune the complexity and obtain an autonomous performance of authoring tasks. Structured guidance and feedback to the authors in the complex scale of the design process is also needed.

Our framework is the basis for the development of such an authoring tool. Our ultimate aim is to attain seemingly conflicting goals: to make an authoring tool, which is powerful, generic and easy to use. The power of the authoring tool comes from the fact that we choose an ontology-based approach. The generality is achieved with the help of defining a meta-authoring tool, which we further instantiate with the concrete learning context to achieve also the power of a domain specific tool. The ease of use comes from the combination of the previous two. As a whole, we are thus able to offer a truly "intelligent" authoring environment. On the one hand we specify reference architecture for IES authoring systems through a meta-level authoring behaviour and a set of general components and their interactions. On the other hand, we allow instantiation of the meta-level framework with specific content and needs for individual IES.

The most common situation with existing authoring systems is that they are either powerful for the purposes they want to achieve in a specific domain (thus very domain-specific) but hard to use and implementing only a particular type of learning, or they are easier to use, more general but thus less powerful for the various domains. The secret to realize our, more ambitious goal is to get in-depth understanding of what an authoring process is by using the concept of ontological engineering that enables us to make computers understand what we understand. The concrete key research contribution is the building of the Authoring Task Ontology (ATO) as part of our authoring environment. ATO enables us to build a meta-authoring tool, which specifies and controls the process of tuning an authoring tool to become configured appropriately for a specific authoring process. Such a meta-authoring tool works analogously to an authoring tool when it generates a concrete learning support system. In other words, an authoring tool configures the components in the target IES according to some constraints such that the IES is built in the same way as the meta-authoring tool configures components in the target authoring tool under the constraints included in ATO. With the help of the ontology it is thus possible to tailor the general architecture to the needs of each individual system to be authored. The main benefit of the use of the ontology is that we can specify the entire authoring process on a meta-level, including both static (concepts, context, and user) and dynamic (behaviour) knowledge in comparable and unified form, so that we can further make it concrete to each specific system. As a consequence, an authoring system for an intelligent tutoring systems (ITS) could be based on an ITS ontology, an authoring tool for group learning courseware could be based on an Computer-Supported Collaborative Learning (CSCL) ontology, and so on.

The remainder of this article contains our views on different aspects of this authoring framework. In the next section, we discuss the main building blocks of IES. Then address issues of authoring IES and the role that ontologies can play in this respect. The authoring support framework is presented, before we focus on the Authoring Task Ontology (ATO) in more detail in the last section where we give examples of how concepts, authoring tasks, and system responses are managed in the ATO. Finally, before coming to the conclusion we address some architectural issues.


Currently there is no agreed upon or standardized reference architecture for intelligent educational systems. However, many of those knowledge-based systems share common architecture features, which can form the basis for an IES reference framework. Such reference architecture will allow for a more structured and unified process of building IES. With it we would also be able to compare, analyze, and assess various IES. The provision of user-oriented instruction depends highly on the following factors:

* The IES is related to a specific subject domain, thus relies heavily on maintaining a Domain Model, describing the structure of the information content within the system (based on concepts [domain terms] and the links between them).

* A major focus in IES is the user and the adaptation of system behavior to the user's personal, cognitive, and knowledge characteristics. Thus, the IES maintains a User Model and User Profile that reflect the user's preferences, knowledge, goals, activity history, current state, cognitive state, and other relevant aspects from an instructional point of view (often as an overlay model of the domain model).

* Each IES also aims at providing an adequate instructional guidance, coaching, and teaching, which is adapted to the user preferences, current status, and knowledge level (presented in the UM) within the stated domain model. Thus, a presence of an Instructional Strategy Model and various Learning Strategies is crucial.

* The IES is a personalized portal for the user to a large collection of learning/teaching material, which also needs to be managed and maintained within the IES. A domain concept-based approach towards organizing this collection of these materials is usually the most common solution.

* Finally, each IES has a specific module to produce and run the prepared educational structure or sequence.

Although the field of ITS and intelligent and adaptive courseware is constantly developing and applying new strategies, still there is a deep conceptual gap between the authoring systems and the authors. The authoring tools (if they exist at all) are neither intelligent nor particularly user-friendly as Jin, Ikeda, Mizoguchi, and Takaoka, (1997) observed. The development process is burdened by the lack of possibility to share or reuse knowledge or other components between systems. The gap between domain knowledge organization and the educational task strategies prevents dynamic adaptation of educational structures (Ikeda, Seta, & Mizoguchi, 1997; Aroyo & Mizoguchi, 2003). Thus, a vital issue to make IES more usable and widely spread, is to decrease the effort in terms of development time and allowing for rapid prototyping of intelligent designs with quick design and evaluation cycles. The making of knowledge-based courseware often requires programming skills for the authors to specify their design choices (Murray 1999). As a consequence, authoring systems should help the author articulate and organize his/her own domain, resource and task (pedagogical) knowledge, and define their own IES skeleton in user-friendly primitives and high-level concepts.

Another important feature of authoring tools is to let the authors see the dynamic behaviour of their IES so that the authors can examine its validity. One of the biggest benefits of having an authoring system is that it can guide the authors and make them aware of the learning goals, of the assessment and instructional strategies, of the links between the learning material, the subject domain, the learning process and strategy, and the learning goals that need to be achieved. With feedback, hints and recommendations that the authoring systems produce (Aroyo, Dicheva, & Cristea, 2002), they help the author through the whole process, show author's misconceptions, and make him aware of other instructional, assessment, and sequencing alternatives. The authoring systems thus help the author externalise his general domain knowledge into a well-structured instructional sequence of goals, learning objects, learning activities, and assessment activities.

Authoring systems also help authors consistent in building their courseware and in preparing their instructions. They monitor the overall process and discover inconsistencies and conflicting situations (the process of building intelligent courseware is typically so complex that a single author is not able to keep it under control by himself). The authoring systems can take over some of the trivial computations to speed up the process. Finally, by modularising the authoring systems, we also achieve a higher rate of reusability such that the different authors or authoring systems can share and interchange data, components, and functional elements.


Now that we have considered the structure of IES and the more general aspects in authoring them, we turn to the specific aspects of the IES authoring process and the application of ontologies. The main motivation of our work is to define a common authoring approach to IES based on existing open and educational standards and a common reference architecture. We are targeting a systematic, structured, unified, and assessable approach to support the entire life cycle of IES. In this context we consider the maintenance and upgrade of learning resources and the conceptualisation of the domain and its relationships to various educational aspects, such as course sequencing and instructional strategies. We establish this link on the basis of modelling the user knowledge, cognitive and personal characteristics, and maintaining a separate learning goal module. The aim is to support the authors in all the phases of the development process working at various levels of detail and abstraction. This will also help to improve different aspects of the quality of intelligent educational systems themselves, such as reliability, support for multiple authors, analysis, comparison, and maintenance.

To structure the complexity and all the aspects of the entire authoring process, we have identified three main groups of IES authoring activities (Aroyo & Mizoguchi, 2003) and extended them to the following list:

1. to model the subject domain (as a representation of the subject domain knowledge, both in terms of what is considered correct by the tutor and what is the most common incorrect knowledge);

2. to annotate, maintain, update and create learning objects, their structure and metadata;

3. to define the learning goals;

4. to select and apply instructional strategies for both individual and group learning;

5. to select and apply assessment strategies for both individual and group learning;

6. to specify a user (learner) model and additional user characteristics and preferences in a user profile and the way they are both used by the IES (e.g. how the values of the user model/profile are used for sequencing, adaptation of tutoring, etc.); and

7. to specify learning/teaching sequence(s) out of learning and testing activities on three levels of abstraction: (a) general, (b) conceptual and (c) concrete learning content levels.

An efficient approach to the realization of these authoring tasks is to employ modelling techniques to construct models of the knowledge that is relevant for the author to perform these tasks. We capture all the processes related to the previously mentioned authoring activities in corresponding authoring modules, such as the Domain Model Authoring, Learning Objects (resources) Management, Instructional Strategies, Assessment Strategies Authoring, Simulated User Model, Course (Learning) Sequencing, and Learning Goal Authoring modules (see Figure 4 later). A common knowledge representation technique (Chandrasekaran, Josephson, & Benjamins, 1999; Guarino & Poli, 1995) offers us ontologies as a way to conceptualise this authoring knowledge in IES. Corresponding ontologies (e.g., Domain Modelling Ontology, Instructional Strategy Ontology, Learning Goal Ontology, Test Generation Ontology, Resource Management Ontology, User Model Ontology) are defined in order to represent the knowledge and to describe the important concepts in each of those authoring modules. They make it easier to provide effective authoring feedback and to analyze the entire authoring process. It also allows the author to be more aware of the goals and the activities he assigns to reach those goals. The main benefit of using such representation is that it implies a clear separation between the content units (IES learning material and domain and instructional specific aspects) and the corresponding authoring activities. This way we allow for more flexible linking between the concrete IES aspects and the general authoring activities. Ontologies, or more precise conceptualising the authoring knowledge in ontologies, are not only a powerful vehicle for achieving efficient management of authoring activities. They also allow for a more standardized approach towards each of the authoring modules and the overall architecture. They provide a broadly agreed vocabulary for the authoring process throughout various types of IES. Moreover, ontologies are the first step towards reusability, sharability, and interoperability of content, knowledge, and functional components.

In correspondence with Mizoguchi and Bourdeau (2000) the knowledge that we capture for the authoring process and that we thus supply to the authoring tool is split in two layers: (a) a dynamic part where a specification of the educational system functionality is given, and (b) a static part where a specification of the authoring and design process is given in compliance with instructional design and theories. This way, we provide the authoring tool with an understanding of how the instructional system works, and allow it to reason on a level of instructional theories and design. With the help of these two layers the authoring support framework serves in correspondence with the three ontology levels introduced by Mizoguchi and Bourdeau (2000). This means that level 1 provides a set of terms as a well-structured shared vocabulary in order to specify the instructional functionality. Level 2 describes the relationships between the terms and defines semantic constraints in the form of axioms. This level builds the major part of the intelligence within the authoring tool. Finally, in level 3 we describe the meta-model of the authoring process interrelated with a systematization of instructional theories. This way, the ontology appears to be a suitable solution for knowledge systematisation within authoring support tools.

Requirements for IES Authoring Support

Considering the earlier mentioned functionality for IES, we obtain the following "requirements" for IES authoring support, as we capture them in our IES authoring support framework:

* authoring systems help authors be consistent in building their courseware and preparing their instructions;

* authoring systems can monitor the process and discover inconsistencies and conflicting situations; the process of building intelligent courseware is so complex that a single author is not able to keep it under control by himself;

* authoring systems can take over the trivial computations to speed up the process;

* by modularising the authoring systems we achieve a higher rate of reusability, such that the different authors or authoring systems can share and interchange data, components and functional elements;

* authoring systems give feedback, hints, and recommendations to the author about the authoring process, his misconceptions and possible ways of making it more effective;

* authoring systems guide the author and make him/her aware of the learning goals, of the assessment and instructional strategies, of the links between the learning material, the subject domain, the learning process and strategy, and the learning goals which need to be achieved;

* with the feedback, hint, recommendations, or guiding of the authoring systems we make the author aware of his/her goal, and help him/her externalise his/her general domain knowledge into a well structured instructional sequence of goals, learning objects, learning activities, and assessment activities;

* the system allows flexibility for the author in terms of being able to accept or reject the system's proposals, but the system should still be able to respond adequately to the further authoring activities, even after a rejection of the recommendation;

* the system offers three levels of authoring: (a) expert mode with a low level of support and no guidance, (b) semi-supported mode with a medium level of support to the author, in the form of moderate suggestions, hints and guidance, and automatically performing computational and reasoning tasks, and (c) fully guided mode with a high level of support to the author, in the form of strong guidance and wizard-like performing of tasks; and

* the system is able to recognize situations and relationships between them and thus recommend to the author based on what other authors have already done in this situation.


To meet these requirements, ontologies play a central role in our framework. On the basis of the layered approach that we explained in the previous section, the challenge of our ontology-based approach is to effectively manage the different ontologies in the authoring process. For this purpose we consider a meta-level evolutional authoring system that facilitates this ontology-based approach:

* We distinguish a meta-level ontology that specifies the processes in a generic and reusable form and allows this way to support authoring of various instructional strategies and different types of learning support courseware.

* We instantiate the meta-ontology with the specific context of the desired educational system we want to build. In other words, on the meta-level we have selected, which components and, which ontologies are needed to build our desired IES and on this instance level we feed those components with the real domain and context of the system.

Thus we use the meta-level to achieve generality and to support analysing the process on the higher level of abstraction, but to the authors we give an easy-to-use system at an instance level. The ontology at the meta-level is the Authoring Task Ontology (ATO). In Figure 1 where we illustrate the phases in IES authoring we see the central role of ATO.

Figure 1 shows the three layers of abstraction in the process of producing an IES. On the highest (meta) level we are dealing with the construction of the generic authoring task ontology (ATO) and implementing it as a main knowledge component in a meta-authoring system. The ATO uses top-level concepts from the behaviour and task ontologies (e.g., domain modelling task ontology, resource management task ontology, user modelling task ontology, assessment task ontology, and instructional task ontology). This way it defines in a conceptual structure the entire authoring process. A repository of domain-independent authoring components is also defined at this level in terms of authoring task templates to be instantiated further on the lower level. Within the second layer, the instantiation of this meta-schema with the concepts, models, and behaviour of an actual authoring context is performed. Here the actual authoring system is set up in order to produce later, the actual IES at the third layer.


Our ontology-based authoring approach allows, through the meta-level ontology ATO, to author at the instance level such that easily various IES can be produced. To achieve this we distinguish the following characteristic aspects of our ATO approach:

* we propose a set of primitive authoring tasks and primitive system tasks, inspired by the notion of generic tasks (Chandrasekaran, 1986), which we use further to achieve more complex composite tasks; those tasks are central elements in ATO;

* we propose a taxonomy of the system responses to the author or to support an authoring activity; this is also a central part of ATO;

* the approach is goal-centred: for example the goal of learning/teaching is central to the architecture of the authoring system; therefore the goal, and the processes around, is separated from the teaching strategy, which means that we give the flexibility to achieve the same learning/teaching goal by applying different strategies; additionally, when the Goal & Goals Processing Module is separated, we can plug in and out different Teaching and Learning Strategies Modules, while our basic common understanding in the system remains intact; this is also an example of our support of the modularisation of the knowledge (authoring system knowledge) which is a key issue in our research;

* we allow three levels of generalization-specification in the course sequence: (a) a general sequence/structure of topics/modules; (b) for each topic we can assign a set of domain concepts, which can describe it--here we also have a sequence within a topic and within the entire structure of topics; (c) for each concept (set of concepts) we can assign learning objects--here we have a sequence of learning objects for each concept and for the entire structure of topics

* when considering the user we look at his/her personal characteristics, at his/her knowledge capacity and at his/her cognitive skills; and

* when we consider learning/teaching strategies we split them in domain-related (dependent) knowledge and cognitive (learning) skills, which are domain independent.

With this architecture we offer a high level of reusability and sharability of knowledge, content, materials, and functional objects between different courseware and authoring systems. The ontology-based approach, the modularization of knowledge (modular architecture), the use of existing web, semantic web, and educational standards (such as SCORM (SCORM 1.2, 2003; SCORM 1.3, 2003) for the sequencing and the interaction between the authoring modules or LOM (LOM, 2003) for the annotation metadata for the learning resources), the encapsulation of processes with the help of web services standards and protocols, and a common specification language OWL (OWL, 2003), RDF (RDF, 2003), and XML, allow this reusability and sharability.

We aim for an authoring environment, which provides a general architecture of the authoring process, and is able to be instantiated and tuned for the purposes of each author and each intended intelligent system to be authored. In other words, we propose an authoring process specification in the terms of an authoring ontology, to be able to further instantiate this ontology for the specific needs and purposes of individual intelligent systems. On the one hand we offer an authoring environment, and we specify the behaviour of the authoring system, and on the other we allow for instantiation of this environment for the purposes of various IES. Finally, we construct the framework in such a way that we obtain an evolutional (self-evolving) authoring system, which will be able to reason over its own behaviour and based on statistical and other intelligent computations will be able to add new rules or change existing ones in the different parts of the authoring process.


We perceive the authoring process as a collection of various authoring tasks over various objects (e.g., domain objects, learning objects, testing objects), where the process of their sequencing and combination is guided by axioms and constraints provided by instructional design. Each object carries a specific role within the corresponding authoring task. The authoring tasks are independent of the IES domain, the educational strategy, and the educational/learning goal. The authoring task ontology (ATO) is based on the notion of "task ontology" (Fukuhara, Kimura, Kohama, & Nakamura, 1995), which serves as a shared vocabulary to describe problem-solving structures of all existing tasks domain-independently (Jin, Ikeda, Mizoguchi, & Takaoka, 1997). ATO is a meta-level ontology, which contains the upper level concepts of the specific IES authoring ontologies. Its role in an authoring environment is to provide a friendly authoring interface, support the verification of the authoring activities and to allow the authoring system to be reusable. In other words, its ultimate goal is to provide a vocabulary for building a model of human problem-solving processes (Jin et al.). We consider an authoring task as the method to satisfy a problem-solving goal. We apply this method by performing a number of authoring activities, which are independent of the domain. We first translate the knowledge of fundamental characteristics of an IES (the IES behaviour) into a "task ontology" and then conceptualise it as an authoring task ontology and finally integrate it as knowledge component in the intelligent authoring architecture.

We have chosen not to describe the complete ATO specification here, but in the following sections we will describe the essence of ATO considering some of its more illustrative parts. We start in the next section with an explanation and examples of the basic ATO concepts. Stepping on this basis, going further, we define the primitive tasks within ATO. We then introduce ways of building higher-level authoring tasks by applying composite relationships on the ATO primitive tasks. As the system response, in the form of feedback or some automatic performance of authoring functions, is a critical component for an 'intelligent' authoring system, we capture main classes of system responses also in ATO.

ATO Basic Concepts

An important illustration of the ATO is the definition of the basic ATO concepts used further in the formulation of the authoring tasks. We build upon the authoring concepts introduced by (Mizoguchi et al., 1997) for a scheduling task: (a) generic nouns reflecting the roles of the objects in the authoring process, (b) generic verbs representing authoring activities over the objects, (c) generic adjectives representing the modifications of the objects and (d) other authoring task specific concepts. We extend this set and make it IES domain-specialized. Following are some examples of instances of these authoring concepts:

* Nouns define the basic concepts for the authoring system to perform actions on. Examples of nouns are: Concept, Topic, Course, Lesson, Module, Domain, Structure, Sequence, Learning_Activity, Assessment_Activity, Resource, Learning_Object, Author, Student, Text, Tool, Media, Metadata, Relationship, Goal, Constraints, Strategy, Cognitive_State, Instructional_Strategy, and so forth.

* Verbs are applied in a combination or sequence of activities with specific objects or concepts and their modifications. They are also defined as a set of procedures representing its operational meaning. Examples of verbs are: Assign, View, Compute, Compare, Modify, Edit, Update, Select, Check, List, Apply, Design, Specify, Execute, Request, Test, Sequence, Return, and so forth.

* Adjectives are applied for modification and identification of objects' attributes. Examples of adjectives are: Appropriate, Selected, Shared, Related, Finished, Required, Requested, Idle, In-Use, Updated, and so forth.

* Other concepts: Activity_Output, Concept_Prerequisite, Lesson_Constraint, Constraint_Satisfaction, Attribute, Predicate, Knowledge, Composition_Operator, and so forth.

Primitive Authoring and System Tasks

Another illustrative aspect of ATO is related to the task definition. The ATO describes the relations among the authoring tasks and the roles of the objects which they play in a particular task. Each authoring task is defined by: (a) sequence of primitive tasks with their activity type, constraints, and input/output resources; (b) goal; (c) requirements; (d) constraints, which can be flexible or hard constraints.

We define primitive tasks, respectively authoring tasks and system tasks, over objects (e.g., domain and course concepts, topics, learning objects, user model and user profile attributes, cognitive characteristics, learning goal) within a specific structure in the authoring system, such as domain model, user model, user profile, course sequence/structure, or learning goal representation hierarchy. Those primitive tasks constitute a basic functional formalism that expresses how the object changes the structure, or the structure is manipulated. Table 1 presents some examples of primitive authoring and system tasks as captured in ATO.

Composite Authoring and System Tasks

At this stage we define a hierarchy of higher-level (composite) functions to represent conceptual categories of relationships (interdependence) between primitive functions. These relationships present certain aggregation criteria (including causal and other relations among components) that are used for grouping primitive tasks into higher-level classes of authoring and system tasks. This way we can construct/identify functional groups of authoring tasks. The higher-level tasks represent a role of one base function for another base function. They are not concerned with the actual change in the objects, but with their actual function in the process of authoring IES. We define those tasks with conditions for their primitive parameters to be able to achieve specific authoring goals. For this we extract the task structure of IES authoring and authoring system behaviour from the previously defined authoring modules previously discussed.

We identify several groups of authoring tasks, related to the domain model construction, simulation of user model, user profile and cognitive aspects of the user, application, and selection of instructional strategy, assessment strategy, and learning goal specification. Within these we define a hierarchical organisation of concepts linked by the ontological link types "is-a," "part-of" (p/o), and "attribute-of" (a/o) (Kozaki, Kitamura, Ikeda, & Mizoguchi, 2002). An example fragment of the ATO is given in Figure 2.

Authoring System Responses

Another part of ATO considers the classification of types of possible system responses. We identify three main groups of system responses.


First, the system can post a message of a specific type. This system response is of a proactive nature and is the resulting reaction to a specific author's action. The message types are identified as one of the following:

* Affirmative The response can be of an affirmative nature that is related to the cause of the reaction: the why. This can be either explicit or implicit:

-- Explicit:

-- Explanation = description of a situation

-- Recommendation = selection of an option

-- Implicit:

-- Hints = targets an understanding of a situation (awareness of the author)

* Notifying The response can be related to notifying the user of what happened:

-- Alerting notification

-- Neutral notification

* Composite Response can be composed out of other responses:

-- Warning = alerting notification + explanation

-- Advise = notification + recommendation (+ explanation)

Second, the system can perform a computation of a specific type. This is an internal system response, which generates input for proactive or reactive responses and usually precedes them. The automatic computation can be in one of the following formats:

* Direct (internal, effective change)

-- Calculations, Statistics, Data mining, Reasoning

-- Update

* Indirect (external, information-oriented)

-- View, List

-- Searching

-- Sorting, Ranking

Third, the system can pose a question of a specific type. This system response could be either of a proactive or reactive nature and is usually performed when there is need for more input or clarification. The question type is not considered so much relevant to the designing of authoring behavior and system responses. The type of the questions is related more to the actual content and the current instructional goal and course task.


To realize the notion of ATO and to gain the biggest benefit from it,

we place our ontology-based framework in a modular architecture. Here ontologies describe the vocabulary and the functionality of each module and a simple standardized API can be provided for the communication between the different modules. While currently most of our research is devoted to the complete construction of the ATO, in this section we want to shortly address some aspects of the assumed tool support for ATO. We illustrate our ideas within the architecture of AIMS--an example of a learning support system (Aroyo & Dicheva, 2001).

AIMS architecture is a good example of applying the principles of the ATO-based authoring support framework. AIMS combines the use of ontologies with educational metadata standards to achieve openness and interoperability of intelligent instructional software. The AIMS authoring model is based on a strict separation and independency of the roles of the domain expert and the course author, which implies separate definitions of the subject domain knowledge and the instructional knowledge (Figure 3)

In compliance with the definition of our ATO-based framework, we identify four groups of authoring activities (Aroyo & Mizoguchi, 2003), namely, (a) domain-related authoring activities supporting the construction (editing and annotating) of a domain model (DM) in terms of concepts and links; (b) course-related authoring activities, involved in the generating of the course sequence and structure; (c) resource-related authoring activities, which are related to the definition of educational resources collection; and (d) user-related authoring activities, with deal with the definition of UM attributes and their simulated application in course task sequencing. Currently, they are realized as a subset of the course-related activities. Thus, AIMS authoring model supports authoring typical for the three main modules of an Intelligent Educational System--Domain Editing, Course Sequencing, and Resource Management. It uses ontologies to describe the authoring knowledge related to those modules. Further, AIMS applies the notion of ATO by capturing the top-level concepts of those ontologies in a meta-level ontology.


One way of ensuring modularity is to split the IES authoring knowledge in several independent parts. On the most general level we distinguish between static and dynamic knowledge to separate between the specification of the authoring and design process in compliance with instructional design and theories, and the specification of the educational system functionality (Figure 4).

The former allows the authoring tool to reason on a higher level of curriculum and instructional theories and design. The latter helps the authoring tool understand how the instructional system works (e.g., knowledge about the tutoring strategy and the learner's adaptation). This way we can achieve separation of the general educational principle (knowledge) and the specific instructional content and processes. To achieve separation of data (content), application (educational strategy), the instructional goals and the assessment activities, we take a goal-centred approach, where a learning goal ontology (and module) is separated from the knowledge on instructional strategies and course sequencing. This allows high reusability of the rather general knowledge on instructional design and strategies. Thus we have a clear distinction between the content and the computational knowledge, where the learning goal plays a connecting role to bring them together within the specific context of each IES.



Our aim in this research is to specify and implement a general authoring framework for content and knowledge engineering for Intelligent Educational Systems (IES). We are developing an authoring tool supporting this framework that is powerful in its functionality, generic in its support of instructional strategies and user-friendly in its interaction with the author. It is an ontology-based authoring environment, since we see the use of ontologies as effective means to have structured content and knowledge engineering. The key contribution of our research is the meta-level Authoring Task Ontology (ATO) specifying authoring tasks, goals, and activities, and knowledge about the process of engineering IES. It enables us to build a meta-authoring tool that defines and controls the configuration and tuning process of an authoring tool for a specific authoring process. Our approach allows such a meta-authoring tool to work analogously to an authoring tool generating a concrete learning support system.

The main added value of this approach is that on the one hand the ontologies in it make the authoring knowledge explicit, which improves the basis for sharing and reusing. On the other hand, it is configurable through an evolutional approach. Finally, this knowledge is implementable, since all higher-level (meta-level) constructs are expressed with a limited class of generic primitives out of lower-level constructs. Thus, we set the ground for a new generation of evolutional authoring systems, which meet the high requirements for flexibility, user-friendliness and efficiency in maintainability.

The ontologies provide common vocabularies and common understanding of the entire IES authoring processes. They allow for interoperation between different applications and authors. This is an important benefit, as currently there are many standards for parts of the problem. The modern (web) society does not go well with a centralized approach (e.g., providing only one educational standard), thus we need to facilitate the integration and extensibility of the knowledge models.

Providing common, meta-level authoring task ontology we can support the common reasoning over the processes accruing in IES authoring. This can be further applied in the reasoning strategies of the authoring support tools. As a result of the analysis and reasoning over the authoring process, the authoring support system will be able to perform various actions automatically and provide hints and recommendations to the author. The ontology-based authoring framework allows making explicit the usually implicit steps/results of the designer's work, so that he can reflect better on the design process (Hayashi, Ikeda, & Mizoguchi, 2004). Further, ATO serves for the semantic integration of all the authoring system components. In our attempt to achieve high levels of modularisation and independence between general and domain specific parts, we consider as future research to extend the current architecture with more detailed specification of the instructional strategies module, by splitting it in designated modules to handle assessment (Soldatova & Mizoguchi, 2003) and group learning strategies, in particular CSCL strategies (Inaba, Supnithi, Ikeda, Mizoguchi, & Toyoda, 2000).
Table 1 Primitive Authoring and System Tasks

Set of primitive authoring tasks

* Create -- Domain_Model, Course_Structure, Learning_Object
* Define -- Composite_Learning_Activity, Course_Sequence,
 Relationship_Type, Testing_Activity, Concept_Relationship
* Add -- Domain_Concept, Course_Topic, Course_Concept
* Delete -- Domain_Concept, Course_Topic, Course_Concept,
 Composite_Learning_Activity, Course_Sequence, Relationship_Type,
 Test_Activity, Concept_Relationship, Course_Structure,
 Learning_Object, Domain_Model
* Edit -- Concept_Attributes, Relationship_Attributes, Metadata,
 Test_Properties, Learning_Activity_Attributes,
 Test_Activity_Attributes, Course_Properties, Sequence_Properties,
 Prerequisites, Sequence_Rules
* Select -- Sequence_Operator, Rule, Domain_Model, Course,
 Domain_Concept, Course_Topic, Course_Concept,
 Composite_Learning_Activity, Course_ Sequence, Relationship_Type,
 Testing_Activity, Concept_Relationship, Learning_Object, Prerequisite,
* Apply -- Sequence_Operator, Rule, Search
* Assign -- Learning_Activity, Test_Activity, Learning_Object
* Relate -- domain_concepts, course_concepts, course_topics,
 Learning_Object, Learning_Activities, Testing_Activities
* Fill_in -- attribute_value, weight_value, metadata
* Execute -- action, message or automatic computation
* Request, Search

Set of primitive system tasks

* Update -- Domain_Model, User_Model, Course_Structure, Course_Sequence,
 Lists, Views
* View -- Domain_Concept, Course_Topic, Course_Concept,
 Composite_Learning_Activity, Course_Sequence, Relationship_Type,
 Testing_Activity, Concept_Relationship, Course_Structure,
 Learning_Object, Domain_Model, User_Model, Concept_Attributes,
 Relationship_Attributes, Metadata, Test_Properties,
 Learning_Activity_Attributes, Test_Activity_Attributes,
 Course_Properties, Sequence_Properties, Prerequisites
* List -- Domain_Concepts, Course_Topics, Course_Concepts,
 Composite_Learning_Activity, Course_Sequences, Relationship_Types,
 Testing_Activities, Concept_Relationships, LOs, Concept_Attributes,
 Relationship_Attributes, Metadata, Learning_Activity_Attributes,
 Test_Activity_Attributes, Prerequisites, Domain_Model, Courses,
 Author_Roles, Sequence_Operators, Rules
* Check -- explicit presence of values
* Identify -- select on a criteria
* Decide -- choose on a criteria and goal
* Request, Compute, Compare, Ask, Respond


This research has been performed in collaboration at Mizoguchi Lab, Osaka University, Japan. The authors are grateful to Akiko Inaba and Larisa Soldatova for their valuable contributions in our scientific discussion.


Aroyo, L., & Dicheva, D. (2001). AIMS: Learning and teaching support for WWW-based education. International Journal for Continuing Engineering Education and Life-long Learning (IJCEELL), 11(1/2), 152-164.

Aroyo, L., & Dicheva, D. (2002, December). Authoring framework for concept-based web information systems. In Proceedings of Concepts and Ontologies in Web-based Educational Systems, ICCE2002, (pp 41-48). Auckland, New Zealand.

Aroyo, L., & Dicheva, D. (2004). Authoring support in concept-based web information systems for educational applications. International Journal for Continuing Engineering Education and Life-long Learning, 14(3), 297-312.

Aroyo, L., Dicheva, D. & Cristea, A. (2002). Ontological support for web courseware authoring. In Proceedings of ITS2002, International Conference on Intelligent Tutoring Systems, June 2002, Biarritz, France, LNCS 2363, Springer, 270-280.

Aroyo, L, & Mizoguchi, R. (2003, July). Authoring support framework for intelligent educational systems. Paper presented at the 11th International Conference on Artificial Intelligence in Education.

Breuker, J., & Bredeweg, B. (1999). Ontological modelling for designing educational systems. Workshop on Ontologies for IES at AIED '99, Le Mans, France.

Brusilovsky, P. (1995). Intelligent tutoring systems for world wide web. In Proceedings of 3rd International World Wide Web Conference (WWW'95). [Online]. Available:

Brusilovsky, P., & Eklund, J. (1998). A study of user model based link annotation in educational hypermedia. Journal of Universal Computer Science (Springer Science), 4(4), 429-448.

Chandrasekaran, B. (1986) Generic tasks for knowledge-based reasoning: The right level of abstraction for knowledge acquisition. IEEE Expert, 1, 23-30.

Chandrasekaran, R., Josephson, J., & Benjamins, V. (1999, January/February). What are ontologies, and why do we need them? IEEE Intelligent Systems, 14(1), 20-26.

Chen, W. Hayashi, Y., Kin, L. Ikeda, M., & Mizoguchi, R. (1998) Ontological issues in an intelligent authoring tool. In T-W. Chan, A. Collins, & J. Lin (Eds.), Proceedings of ICCE'98, 1, 41-50.

De Bra, P., & Calvi, L., AHA! An open adaptive hypermedia architecture. The New Review of Hypermedia and Multimedia, 4, 115-139.

Devedzic, V. (2001). The semantic web--implications for teaching and learning. In Proceedings of ICCE 2001, (pp. 26-28). Seoul, Korea.

Devedzic, V., Jerinic, L., & Radovic, D. (2000). The GET-BITS model of intelligent tutoring systems. Journal of Interactive Learning Research, 11(3), 411-434.

Fukuhara, Y., Kimura, F., Kohama, C., & Nakamura, Y. (1995). A knowledge-based educational environment integrating conceptual knowledge and procedural knowledge in telecom. In Proceedings of ED-MEDIA'95, (pp. 229-234).

Guarino, N., & Poli, R. (Eds.) (1995). Formal ontology in conceptual analysis and knowledge representation. Special issue of the International Journal of Human and Computer Studies, 43(5/6), 625-640.

Hayashi, Y., Ikeda, M., & Mizoguchi, R. (2000). Is what you write what you get?: An operational model of training scenario. In Proceedings of ITS'00, Montreal, Canada.

Ikeda, M., Seta, K., & Mizoguchi, R. (1997). Task ontology makes it easier to use authoring tools. In Proceedings of IJCAI" 97, (pp. 187-190).

Inaba, A., Supnithi, T., Ikeda, M., Mizoguchi, R., & Toyoda, J., (2000). How can we form effective collaborative learning groups--theoretical justification of opportunistic group formation with ontological engineering. In Proceedings of ITS'2000, (pp. 282-291). Montreal, Canada.

Jin, L., Ikeda, M., Mizoguchi, R. & Kitakaoka, Y. (1997). Role explication of simulation in intelligent training systems by training task ontology. In Proceedings of Al-ED-97 (pp. 41-48). Workshop on architectures for intelligent simulation-based learning environments.

Kiyama, M., Ishiuchi, S., Ikeda, K., Tsujimoto, M., Fukuhara, Y. (1997). Authoring methods for the web-based intelligent CAI system CALAT and its application to telecommunications service, (p. 44). AAAI Fall Symposium.

Kozaki, K., Kitamura, Y., Ikeda, M., & Mizoguchi, R. (2002, October). Hozo: An environment for building/using ontologies based on a fundamental consideration of "role" and "relationship." In Proceedings of the 13th International Conference on Knowledge Engineering and Knowledge Management, EKAW 2002, (pp. 213-218).

LOM (2003). LOM IEEE P1484. 12 Learning Object Metadata Working Group. [Online]. Available:

Major, N., & Ainsworth, S. (1997). Developing intelligent tutoring systems using a psychologically motivated authoring environment, (p. 53). AAAI Fall Symposium.

Mizoguchi, R., & Bourdeau, J. (2000). Using ontological engineering to overcome common AI-ED problems. International Journal of AIED. 11(2), 107-121.

Mizoguchi, R., Ikeda, M., Sinitsa, K. (1997). Roles of shared ontology in AI-ED research: Intelligence, conceptualization, standardization, and reusability. In Proceedings of AI-ED 97, (pp. 537-544). Kobe, Japan

Murray, T. (1999). Authoring intelligent tutoring systems: An analysis of the state of the art. International Journal of Artificial Intelligence in Education, 10, 98-129.

Nkambou, R., Gauthier, G., Frasson, C. (1996). CREAM-Tools: An authoring environment for curriculum and course building in an ITS. In Proceedings of Third International Conference, CALISCE'96, 1108, 186-194.

OWL (2003). Web ontology language. [Online]. Available:

RDF (2003). RDF vocabulary description language: RDF schema. [Online]. Available:

Redfield, C. L. (1997). An ITS authoring tools: Experimental advanced instructional design advisor, (pp. 72-82). AAAI Fall Symposium.

Soldatova, L., & Mizoguchi, R. (2003). Ontology of tests. In Proceedings of Computers and Advanced Technology in Education Conference, CATE'03, (pp. 175-180). Rhodos, Greece.

SCORM 1.2 (2003). SCORM 1.2 specification, ADL Initiative. [Online]. Available:

SCORM 1.3 (2003). SCORM 1.3 specification, ADL Initiative, [Online]. Available:

Vasilakos, T., Devedzic, V., Kinshuk, & Pedrycz, W. (2004). Computational intelligence in web-based education. Journal of Interactive Learning Research, 15(4), 299-318. Special Issue.

Vassileva, J. (1995). Dynamic courseware generation: At the cross point of CAL, ITS and authoring. In Proceedings of ICCE'95, (pp. 290-297).


Eindhoven University of Technology, Eindhoven, The Netherlands


Osaka University, Osaka, Japan
COPYRIGHT 2004 Association for the Advancement of Computing in Education (AACE)
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Author:Mizoguchi, Riichiro
Publication:Journal of Interactive Learning Research
Geographic Code:1USA
Date:Dec 22, 2004
Previous Article:A framework for concept-based digital course libraries.
Next Article:Adaptive learning object selection in intelligent learning systems.

Related Articles
Devolution and choice in education: The school, the state and the market.
Michael Fabricant and Robert Fisher, Settlement Houses Under Siege: the Struggle to Sustain Community Organizations in New York City.
Syaukani H.R., 2001, Pelaksanaan desentralisasi di bidang kehutanan dalam rangka pengelolaan hutan lestari di kabupaten Kutai Kertanagara.
Six Modern Plagues and How We Are Causing Them.
Power to the region: GEMSA is born.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters