2.9.1 Separation of concerns
The separation of concerns principle is designed to ensure reusability and flexibility. The principle ensures that (e.g.) adaptation specification is separated from the domain content. The layered structure of the AHAM model (and therefore AHA!), is designed to keep the domain, adaptation and user models separate from each other. Wu et al. [29] noted the need within AHAM for separating the adaptation model from the domain model, however adaptation content was also often stored within the static content. For this reason, the goal and constraints layer added to LAOS [35] was designed to ensure that only the factual domain content is stored in the domain model, and pedagogical information is shifted to the goal model. The separation of concerns is inherited from object-oriented programming [72]. It is also closely related to the ‘separation of content from presentation’ principle that is employed with XHTML2.0 and CSS [73], which allows authors to change the
presentation of a number of web pages by simply redefining the CSS files – thus promoting reuse.
2.9.2 Requirements for an Adaptive Hypermedia Authoring System
Based on this related research, we can identify an initial set of principles that are necessary for an adaptive hypermedia authoring system.
27 DP1. Use of Frameworks. The frameworks described in Section 2.2
demonstrate how the frameworks already implement the separation of concerns principle to some extent. For this reason, it is recommended that any new adaptive (delivery or authoring) system follows a framework. Broadly, the frameworks all specify that there are two different types of content that need to be specified; static content, and adaptive content. a. An AHAS should adhere to a Static Content Specification. As the
above background research has demonstrated, most adaptive hypermedia systems use HTML. However, there are no widely- accepted standards for describing the static content used within adaptive hypermedia. There are, however, many standards (such as IMS-CP, IMS-QTI, IMS-LD3 and SCORM4) that are widely used within e-learning systems. In the absence of any static content standards for adaptation, Ghali [60] created methods of allowing such e- learning standards to be utilised within an adaptive course.
Moreover, Chapter 4 further explores standards and formats that are frequently used for creating educational content.
b. An AHAS should adhere to an Adaptive Content Specfication. Some previous systems (such as Interbook [27] and early versions of AHA! [74]) combined the static content with the adaptation specification. Using some form of specification (such as LAG, or potentially a more
3 http://www.imsglobal.org/
28 generic scripting language) the adaptive content can more easily be separated from the content, and thus be reused. In this context, we consider all user model and presentation model issues to be directly related to the adaptation – and therefore also covered by this principle.
DP2. An AHAS should support reuse. To remove the burden of authoring new content, it is important that the authoring system allows the author to reuse existing content. This requirement was used by Cristea et al. when comparing frameworks in [75] and can be divided into two sub-
requirements.
a. An AHAS should allow linear content import. As Section 2.7
described, many authors will already have prepared linear resources in various kinds of media and file formats for use within other environments – either real-world universities (such as presentation slides) or in virtual learning environments (such as course pages authored in Moodle or Sakai[66]). The adaptive hypermedia authoring system should accommodate file formats that have already been created for other purposes and, where possible, enhance the value of these files via adaptation. Ghali [60] previously researched integration with learning management systems and Chapter 4 details methods of extracting content from other linear sources.
29 b. An AHAS should support the author with reuse. Within an adaptive
system, it is likely that authors will want to reuse educational content that has already been created. The system should provide a way of helping the user to maintain and reuse currently stored content (and searching through it). This could be content that they themselves had previously created, or the author may wish to use content that has been created by another author, thereby allowing a collaborative process such as that used by Baloian et al. [76].
Similarly, it is also important that reuse also applies to the
adaptation strategy, thereby allowing non-technical users to take advantage of the code that has been implemented by programmers. DP3. An AHAS should ensure an acceptable level of usability. To encourage
non-technical content authors to use the AHAS, the tool must be usable. This principle can be divided into two main sections.
a. An AHAS should be consistent with other applications. The authoring system should provide consistency [77] with other well-known applications, especially functions that are common to more generic authoring systems should be easily recognizable to beginner users. b. An AHAS must have a shallow learning curve. The application should
also allow users of all abilities to create complex adaptive content that utilizes all possible benefits of adaptive hypermedia. Such
functionality will need to be presented simply to the user, requiring a shallow learning curve.
30 2.9.3 Description of existing tools with respect to these principles
In this section, three adaptive hypermedia authoring tools are described with respect to the above requirements.
Interbook does not use a particular framework (DP1), and therefore has no way of separating static content from the adaptation rules (DP1.a and DP1.b). There is scope for limited adaption reuse in Interbook, through the use of LISP. Interbook inherits many of its user interface features from Microsoft Word. This means authors can import existing Word documents into the system to use as the basis for their material, and also reuse any documents prepared for adaptive hypermedia (DP2.a and DP2.b). Although much of the interface is Microsoft Word (DP3.a), the author still requires understanding of how to apply adaptation rules (DP3.b).
AHA! uses the AHAM [30] framework, but also allows mixing adaptation rules with static content (DP1.a and DP1.b). Content must be authored in (X)HTML, for which many authoring tools exist (DP2.a and DP2.b). AHA! comes with a tool to import content (to upload it to an AHA! server). Adaptation in AHA! can be defined graphically, without any knowledge of the adaptation rule language (DP3.a and DP3.b). However, authors do need knowledge of the adaptation rule language in order to define new complex adaptation rules.
Due to the combination of the LAOS framework, the HTML content, and the ability to export to CAF, MOT1.0 fulfils requirements DP1, DP1.a and DP1.b. In terms of content reuse, HTML can be used as a basis for static content (DP2.a). MOT1.0 also allows static content to be copied and linked to other static content – this makes
31 reusing information from other domain maps, possible (DP2.b). MOT is a web- based tool, and therefore inherits many user control paradigms from general web design principles. For example, hierarchies in MOT1.0 are rendered using static HTML. However, the interface used in MOT1.0 is inconsistent with other applications, therefore not completely fulfilling requirement DP3.a. Previous evaluations highlighted usability issues within MOT [46], [78] (such as too much complexity in the process of labelling content, and the hierarchy display), therefore highlighting problems with the interface, DP3.b. These usability issues are discussed and addressed in Chapter 3.
Based on these requirements, MOT1.0 demonstrates the most flexibility for authoring. However, the requirements also highlight areas that could be improved within MOT.