If we backup and restore to a new course, that brings up the question: Can we train a model on data from multiple courses knowing that those courses will be very similar but might have some minor differences with each instance of the course?
I am guessing the LA engine is working off the log but does if it depends on a single course in the log, that would say we really need to reset rather than restore as a new instance each semester.
This brings up another question. If we reset the course each semester, and we need to make some changes (new versions of slides for students to download, new links to visit, etc.) how can the LA engine see, or rather how can it represent to the consumer of the model's output, what were the atifacts at the time each student used them?
I am guessing the LA engine is working off the log but does if it depends on a single course in the log, that would say we really need to reset rather than restore as a new instance each semester.
This brings up another question. If we reset the course each semester, and we need to make some changes (new versions of slides for students to download, new links to visit, etc.) how can the LA engine see, or rather how can it represent to the consumer of the model's output, what were the atifacts at the time each student used them?