Hi Timothy,
thanks for the detailled answer.
Here are some comments that may precise things and maybe "reconciliate" some point of vue:
First I don't want to redo Copilote in PHP. I will only use some parts of it internal knowledge representation to support students learning. More on that later.
Thanks for the links. My French is of insufficient power to allow me to read
the copilote explanation in any detail but I get the general drift.
There is an English language page here
http://users.edte.utwente.nl/lanzing/cm_home.htm
which explains the formal / symbolic system upon which copilote was based
(I believe).
Concept mapping was studied by copilote authors but they retains 3 others visual knowledge representations for 3 differents levels of knowledge.( empirical, hiearchical and actantial )
I think that Copilote is not going to be terribly high on Martin's wish list but as
far as I am concerned it looks very useful.
Martin's wish list and mine are not so divergents: we want, and you also I am sure, to give to the learners the best of each teacher capabilities and to support them with an excellent improving tool: Moodle.
IMS QTI is the way vs adaptive feedback
It is sure that my post can sustain a comparision with IMS document.
But these items correspond to that:
2.7 Reduce Content in Learning Path Based upon Learner Profile
2.8 Using Virtual Labs
2.10 Adaptive Learning Delivery
If we want to offer theses additionnal services to the learners then we must do:
1 - Create some import scripts that read xml existing content and that feed Moodle database with that content.
2 - Modify and/or add to Moodle modules to support IMS functionnalities.( linking things, learner profile, ... )
Items 1 and 2 must fit together and may take a long time to develop.
It look like Chicken and the Egg Dilemma
If we start by #1 then the learners will have access to many more imported questions without adaptive feedback. And it will be correct to claim that Moodle is IMS compatible.
If we start by #2 then th learners will have access to some form of adaptive feedback but with only the few questions that the teacher will create.
At the end, after the chicken sit enough long time over the egg, we will have 2 chicken abilited to live together
On my side I don't have any stuff to import from IMS or any other format. If I have some time for work I will choose #2.
But someome migrating from WebCT or Blackboard or any other will appreciate some importing even if all the functionnalities are not supported. Here #1 is the best choice.
I have still to make enough feedbackless quizes (I try to include feedback)
I think it will be some time before I manage to include more complex feedback.
Here you abord the worst side of LMS or IMS or Moodle. More you add fucntionnalities to your system more it is complex and time consuming for the teacher. The f=ma example with only 5 choices will require 25 other choices if we give 1 level adaptive feedback and 125 choices for 2 levels adaptive feedback. If you use 10 randoms questions then 1250 choices. Crazy with currents authoring tools!
Here come the reason for my reference to Copilote. To simplify it, Copilote internal knowledge representation ( not it visual presentation ) is base on some form of Bloom's Taxonomy with about 11 levels. Copilote software only store these levels. It is not capable to shift from one level to the other alone. My contribution will be to program interlevel correspondance to support students.
Here is an simplified example: suppose that a student have some problem to sort somethings. Maybe he don't know on what propriety to sort, what are theses things or what is sorting! The system will be able to downshift one level from sorting and with generated adapted choices will try to evaluate what the student is missing. After the software will try to give simple examples of what the student is missing. If the software reach the lowest level that the teacher has entered, then it will refer the student to the teacher for an intelligent action. My 200 english words vocabulary are not sufficients to explain that clearly but I hope that 2 or 3 weeks of work later will make a working prototype that will permit to experiment that with some ( not all ) levels programmed.
But this is something that would be shareable and reusable. I am sure that there are
going to be more and more systems like this and that they will be very useful especially
in language education -- or the study of large formal systems with a fairly
universally accepted system of rules, also physics as you pointed out.
Another relevent thread on Re-usable Adaptive Educational e-Content,
mentioning AIML and Martin's reaction can be seen at
http://moodle.org/mod/forum/discuss.php?d=1235#4995
For the shareable and reusable sides, from the precedent example, suppose that an mathematic teacher do some work on numbers system, then his work may be linked to the missing level of the example above. ( suppose that you whant that your students sort Austria, Japan and Quebec by population and by superficies. In your feedback you don't have to explain what is a superficie. If someone is ok for population but bad for superficies the system will refer it to you. If you find a good link of a math teacher about superficies in the system, you will link it in your feedback. )
Thanks for your feedback,
Bernard