Project

General

Profile

Standardize Content & Domain Models in ITS: A few ideas -- Graesser & Hu (from U. of Memphis)

Added by Hu, Xiangen about 1 year ago

This is from Graesser & Hu (of U. of Memphis)

Abstract: When considering standardization for each of the four components in ITS, domain models may be one of the better understood components compared with others. It is content and domain modeling that most subject matter experts think about when they create curricula and learning environments. Stated most simply: “Content is king.”

There are already widely used models for characterizing domains, such as “concept maps” (http://cmap.ihmc.us/docs/conceptmap.php), Knowledge Space (Learning Space) (http://liinwww.ira.uka.de/bibliography/Ai/knowledge.spaces.html), semantic taxonomic networks (such as the semantic web ontologies), and organized scenarios. When we consider standardization of a domain model for ITS, we might consider the following:

  • It should reflect the natural way that content developers think about the domain, as opposed to obtuse and misaligned with their thinking.
  • It needs to be mathematically and computationally specifiable so that it is feasible to implement.
  • It should be consistent with existing ways of specifying domains, including those mentioned above. All of these are structural and relational. There are standard ways to specify types of items items and relations between items, for example. There should be several levels, aspects, and grain sizes. That would be standardized specification of:
    1. the structure of the domains,
    2. the elements (singletons or groups),
    3. relations between the elements,
    4. packages of content (e.g., scenarios, schemas),
    5. formal and quantitative assumptions, and
    6. links to the other components of standardization.

Replies (7)

Avatar?id=10&size=24x24 RE: Standardize Content & Domain Models in ITS: A few ideas -- Graesser & Hu (from U. of Memphis) - Added by Sottilare, Robert about 1 year ago

Is there really going to be an opportunity to standardize domain aspects or are things changing too fast to make standards practical? (Barr)

How far do are you thinking about breaking down domains? cognitive, psychomotor, social? or more granularity? (Sottilare)

I think the structure and constraints might be the level at which we might breakdown various domains for use on a variety of computer platforms. (Sottilare)

RE: Standardize Content & Domain Models in ITS: A few ideas -- Graesser & Hu (from U. of Memphis) - Added by Robson, Robby about 1 year ago

The issue I see, which Paula's ADL experiment points out, is separating what I would call "competencies" (KCs, Learning Objectives, included) from "content" and the relationships among them. Several standards exist for representing domain competencies and also for metadata that describes content. But when you get down to the granularity used by ITS domain models, competencies and content tend to become conflated. The right way to go might be simply to "standardize" how to separate these, using existing standards, if that makes sense.

The other issue is that none of the standards do an adequate job of defining relationships. On the competency side, for example, the standards can express that two KCs are the same or similar or that one is required by the other, but I believe there are many tacit relationships used by actual systems in which relationships are context dependent, e.g. the relationship between one content object or one competency and another depends on learner history. It might be useful to look at whether there are applicable standards for context-dependent relationships. I don't know the answer.

RE: Standardize Content & Domain Models in ITS: A few ideas -- Graesser & Hu (from U. of Memphis) - Added by Durlach, Paula about 1 year ago

I didn't really capture good info on my search for DOD ontology info that I did before, so I started over. And for some reason today I found some more modern hits. Must have used different search terms (?) So I haven't really processed these yet...

If there is some thought for me to do another paper on DOD ontologies, I am willing to try.

http://dodcio.defense.gov/Library/DoD-Architecture-Framework/dodaf20_ontology1/
http://www.ideasgroup.org/
https://www.researchgate.net/publication/224624942_Using_a_Military_Information_Ontology_to_Build_Semantic_Architecture_Models_for_Airspace_Systems

RE: Standardize Content & Domain Models in ITS: A few ideas -- Graesser & Hu (from U. of Memphis) - Added by Durlach, Paula 12 months ago

I spent a lot of time trying to figure out the Dodaf ontology, only to conclude it was not relevant; however it is based on the upper ontology IDEAS, which might be relevant. Basing domain models on upper ontologies may make them extensible but authoring them is not easy.
I also investigated NIEM and thought it might be useful, so have put in another abstract to just describe its capabilities, and suggest that it might be useful for standardizing information exchange between applications (like student data) and within applications (like between different modules, as per Bob's abstract).

RE: Standardize Content & Domain Models in ITS: A few ideas -- Graesser & Hu (from U. of Memphis) - Added by Hu, Xiangen 12 months ago

Enjoying reading all the comments! This will help us to formulate an active discussion topic/session in Orlando. We invite you all be part of the position paper Art and I will be working on.

RE: Standardize Content & Domain Models in ITS: A few ideas -- Graesser & Hu (from U. of Memphis) Added by Vlad Goodkovsky - Added by Goodkovsky, Vlad 11 months ago

Let me add my 2 cents for your consideration:
1. I also tried to find a parent vocabulary/ontology for ITS. No success yet.
2. For developing ITS vocabulary/ontology/domain model, it makes sense to separate declarative/conceptual and procedural domains and focus on the first one.
3. Not limit yourself to bottom-up approach: designing separate parts (domain, learner, pedagogic, interface models) and then trying to put them together. As a rule, putting together requires redesigning parts. It may be a long iteration loop.
4. Try top-down approach: finding a holistic super-model (such as a general activity, control system, ...) and trying to specify it in depth not breaking its consistency and integrity.
5. Take a few existing holistic ITS platforms and try to analyze, reveal commonality, refine, and generalize them in hope to develop standard solutions.
6. Do not mix the domain model (entities, elements, variables, values, relations, ...) and the learner model (competencies, KSA, beliefs, ...)
7. Do not try to standardize specific levels (such as psycho-motor, conceptual, social skills). Good standard models should work on any level.
If all of it make sense of course.

    (1-7/7)