For nearly 20 years CAPDM have been building client content in single sourced, SGML/XML. All CAPDM clients have a digital, managed repository of their content, all of which is held in standards. The textual content uses a very semantically-rich mark-up which offers a high degree of future proofing, but also ensures that their content is agnostic with respect to transient technologies and able to be delivered to all formats and platforms.
At first most clients do not understand the benefits of this approach when it seems easy to slap together a course in Moodle or whatever. However as the technology changes under their feet, or when they go through their nth update of a module the value of this “information management” approach becomes apparent.
Over the years CAPDM clients have pushed us to derive more benefit from this investment in content, which we love clients to do. One particular direction has taken us through Progress Profiling where students can see how their assessment achievement can highlight their level of understanding of assessments based on particular Learning Objectives.
CAPDM always encourages clients to design and develop their content around Learning Objectives because it is potentially possible to do “smart” things with the underlying information and data derived from it. It is never a bad thing to be collecting huge amounts of valuable learning analytics.
For example, if there is a large enough underlying question bank then it is possible to create targeted assessments based on particular Learning Objectives, e.g. those where a student is weak. In the Profile above the peaks show the level of assessment for each learning objective, the black shows what a student could have achieved with what they have done so far, and the green shows what they have actually achieved.
CAPDM have extended this to feed back into the “content” of the course to similarly highlight which areas seem to be well understood and which may be less well understood. This is a first step towards full adaptive learning, though it avoids making very bold claims, e.g. by only showing content still to be mastered. This possible hiding of content, based on a machine-derived conclusion that it is understood, could be dangerous!
What is Adaptive Learning? Adaptive learning is a goal for many, but possibly understood by too few. There are very many papers, reports and even systems that purport to include or offer Adaptive Learning approaches or solutions.
If you are new to Adaptive Learning, or want clarity, then here is a worthwhile paper to read from the DeVry Education Group: DXV White Paper.PDF
This paper has a wonderful Appendix that explains six identifiable levels of adaptivity in learning.
CAPDM’s approach differs from all other providers: we do not push a proprietary system that you have to use. CAPDM’s approach builds on digital repositories of standards-based learning materials that it builds for all clients, but in particular the semantically-rich content marked-up in XML.
CAPDM’s approach to Adaptive Learning is therefore open and extensible, and able to be implemented for a range of platforms. We supply a basic set of rules set interpret the level of learning i.e. mastery achieved by an individual, but this can be replaced with a custom client-produced set if required. The adaptive behaviour can be abstracted from the data of the individual only or from data aggregated from all students.
We call this ALFA – Adaptive Learning For All as it is not restricted to proprietary platforms or “black-box” algorithms that you have no control over. It is also an option, not a mandatory part of an offering within any learning environment.
ALFA does not rely on expensive, proprietary and therefore closed systems. It can be made freely open to every client and course. However never think that good things come for free. While there is no great financial hit to take with ALFA, it can only be as good as the content and metadata associated with that content. Clients need to make that extra bit of design and authoring effort. Again this is not necessarily a huge task, e.g. the information needed for Profiling is very light. Targeted assessments require a sizable question bank to be developed, and a bit more structural work is need to take this to adaptive learning (Level 4 in the paper quoted above).
Finally, it is not all or nothing. Progressing up the layers to adaptive learning can be done in stages, with each extra layer building on the earlier layers. It can be done in stages, if and when the benefits can be clearly identified. ALFA can be layered to provide:
- simple Learning Objective feedback and Profiling
- random, targeted quiz generation
- adaptive pathway construction.
