A Framework for Model-Based Adaptive Training

The Symbol Level

The symbol level is about the way knowledge is represented and implemented. The training software architecture accommodates a set of software tools and facilities capable of realising a design (derived from the mapping of a problem into a training unit design specification). The main components of a realised expert agent are based on the three levels of subject model generality or applicability. In ITSIE, the available knowledge could be delivered by three prototyping tools:

  1. an Event-Graph Language (EGL) for representing procedural knowledge;
  2. a classical Rule-Based Language (RBL) to represent associative knowledge; and
  3. a Component Based Language (CBL) that structures a physical system around physically identifiable sub-systems for representing principled knowledge.

In the MOBAT framework a variety of similar tools are possible.

The models in the software solution for the MOBAT application experiments have shown a close correspondence with the models identified at the knowledge level. This offers more straightforward software design and easier maintenance.

However, it is not always useful to aim for an entirely seamless translation into symbol level code. Depending on the programming language, the software model may differ due to efficiency considerations.

The knowledge level offers an abstract view which has the advantage of eliminating detail with a clear specification of behaviour.

© | | Sitemap