This series of articles has attempted a preliminary assessment of the adequacy of existing e-Learning standards for supporting the introduction of adaptation techniques in e-Learning systems. The analysis, however cursory due to space limitations, has pointed out that existing standards do have some provisions for adaptation, but require substantial extensions to accommodate common practice in ALEs. Our findings can be summarized as follows:
In terms of domain modelling (i.e., modelling of courses, or learning-related activities), existing e-Learning standards do not suffice to capture the rich semantic structure that underlies static learning materials, or single- / multi- participant processes. Several alternative approaches are possible for integrating such semantic-level information into the metadata structures that typically accompany learning materials, as has been clearly demonstrated by recent research efforts. Of these approaches, the ones that seem to be most promising are those that seek to formalise / standardise ways for semantically articulating relationships and properties of the “units” from which materials and activities are composed; alternative adaptation methods and techniques can then be devised on the basis of the available information. More restricted in scope and viability seem to be the approaches that attempt to integrate the adaptation “logic” itself with structural metadata (e.g., course metadata in the case of the KOD project). The main potential drawbacks of the latter approaches are: (a) only the logic that is already incorporated can be applied in adapting / personalising materials, and (b) the incorporated logic is of a procedural nature, and as such of no particular semantic value, making unfeasible its use as input towards the application of alternative adaptation methods / techniques.
The modelling of individual learners as users of an ALE is also not sufficiently covered by what is already available. The “PAPI” set of standards seems to be a step in the right direction, but is not adequate by itself. Not only that, but it is currently “competing” against a host of other related specifications, which may be more likely to adopt by organisations that seek to comply with a series of standards / specifications that come from the same body (e.g., IMS LIP is more likely to be supported by organisations complying with IMS specifications in general). ALE-oriented standardisation (not to mention “de-fragmentation” of learner profiling work) would greatly be facilitated by the convergence of these specifications, with (portions of) PAPI providing explicit support towards adaptation.
The modelling of groups of learners, on the other hand, is still in an embryonic stage in terms of standardisation. Although there exist today ALEs that do identify and represent such groups on the basis of learners’ performance, preferences, interests, goals, etc., the representations used are intended only for internal “consumption”. Although PAPI can be used to convey information about groups of learners, it lacks the expressive capacity to capture the very information that would be of particular interest in ALEs: the common attributes / characteristics of learners (derived, e.g., from their history of interaction with the ALE) that have resulted in their being classified as a group. Specifications that address this shortcoming would have to be developed anew.
Adaptation modelling in the context of ALEs is a more complex issue, because it potentially involves both adaptation “logic” and the “actions” that result from the application of that logic in relation to the various static and dynamic models maintained by an ALE. It has been argued that seeking standardisation at the level of the “logic” might be futile at present, mainly due to the proliferation of approaches that exist currently, and the vast differences between them (e.g., formalism used, computational models, etc.) A more realistic goal might be the standardisation of adaptation actions (i.e., the “things” that can be adapted within an ALE, and the “ways” in which they can be adapted). This would enable the employment of widely differing approaches to “logic”, while unifying, to some degree, the representation of modifications that can be adaptively effected in learning materials, computer support for inter-personal activities, etc. Furthermore, IMS LD has been discussed as a potential standard for instantiating adaptation logic and actions, at a given point in time and for a given user, or group of users. As in the case of group modelling, standardisation efforts in this direction would have little to build upon currently.
Standardisation at the level of adaptation components and services has only recently been addressed (at the level of de facto standards) in the context of research efforts. Two different approaches were presented and discussed in this paper: the extension of specifications that deal with the interchange between components of LMSs, and the introduction of new specifications that are explicitly intended for enabling the exchange of adaptation-oriented information / services between the major parts of an ALE. It is argued that these two approaches are not necessarily contradictory, or, for that matter, mutually exclusive. In fact, they seem to have complementary advantages and drawbacks: the former signifies an only partial departure from existing specifications, but fails when it comes to distinguishing between functionalities that can be exposed as services from specialised sub-systems; the latter, while addressing the problem just mentioned, must be approached very carefully in order to ensure that it does not interfere with, or render unusable, lower-level specifications that are already in wide use. Given the above, it is argued that a combination of the two approaches would be the better grounded alternative for any future endeavours in this direction.
In closing, we would like to touch upon a few topics that we feel are inevitably intertwined with any effort to expand upon current standards and specifications in the direction of adaptation / personalisation. To start with, it is argued that extensions to standards / specifications should happen in a way that keeps the “entry cost” of employing adaptation facilities in the development of e-Learning materials, to as low levels as possible (mainly in terms of invested resources). An example of what would constitute, in the authors’ opinion, a gradual and non-taxing path towards such employment, would be as follows. Authors should be able to provide an existing course with “traditional” metadata to an adaptive system, and get basic adaptation facilities (resulting from a “default” interpretation of the course structure and material by the system). Later on, authors could progressively add “adaptation metadata” as a stepwise approach to enabling / providing more advanced adaptation features.
Secondly, it is important that future extensibility of (new or enhanced) standards is seriously taken into consideration. It can be anticipated that the progressive uptake of adaptive methods and techniques in e-Learning systems will give rise to new adaptation patterns, and an even wider range of approaches than are in use today. Where possible, therefore, new standards / specifications should provide all the necessary extension points that would allow for the progressive enrichment of the respective models.
Finally, the adoption of the new standards or extensions proposed in this paper is, in our opinion, highly dependent upon the development of authoring tools that facilitate the creation of compliant resources. The creation of high quality-, standards compliant- learning material is already a quite demanding goal. The introduction of adaptation facilities will inevitably impose an additional “burden” on content creators. In order to bring the related cost / benefit ratio to non-prohibitive levels, it is necessary to have tools that: can assist authors in converting “static” material; support the authoring of adaptive content; enable the specification of adaptively supported activities in ALEs; etc.
No comments:
Post a Comment