RuleWorks

A Framework for Model-Based Adaptive Training

To select and switch to the appropriate model

The didactic tactician is concerned with training strategy and selecting appropriate models which results in the generation of a particular sequence of training units. In Workmanship-MOBAT, the main tasks for the didactic tactician are to present, to dictate, to coach and to facilitate.

As appropriate during a dynamic training situation, these tasks are used to select and switch to an effective domain model. When a model is selected, it is linked to properties of a relevant training unit. The training units for Workmanship-MOBAT are classified as presentation, example, practice and assessment (see Section 6.3). The four didactic task primitives are presented in greater detail below.

The task primitive “to present” is used for each of the training strategies (e.g., tutoring, coaching, facilitating and discovery). Which domain model is presented depends on the preferred learning mode and expected trainee goal expertise. This means that this task is mapping the rote, inductive or deductive learning modes to the generality of a model.

The trainer can apply “to present” to switch to a new model when a training goal has been completed. The trainer methods available to implement this task are called the propose-example-driven-method, propose-problem-driven-method and propose-test-driven-method. The methods used here are adapted from the ITSIE ‘method library’ [ITSIE 1992, D7].

In ITSIE these methods are related to stored skeleton training plans. In Workmanship-MOBAT, the trainer methods are independent of skeleton training plans. The methods are mapping a preferred learning mode and expected trainee goal expertise level to an appropriate model. At run-time trainer tasks are proposed and selected as needed.

Trainer tasks are combined with trainer methods. Each trainer method starts with identifying possible models depending on the expected trainee goal expertise level. The trainer then presents a sequence of training units which consider the preferred trainee learning mode.

In ITSIE, a fourth method called the “explanation-driven” method is also available. In Proto-MOBAT, this method is not applicable since all trainer methods can make use of an expert’s explanation facility. When the trainee asks for clarification, the behaviour of the expert is justified by the expert’s explanation facility. The expert can justify its underlying reasoning process and answer trainee queries. Each of the three trainer methods that can be combined with the trainer task “to present” is described further with examples below.

  • Using the “propose-example-driven-method” the expert solution to a problem is presented. In Workmanship-MOBAT, this method is appropriate for rote and inductive learning modes. The rote learning mode is used with the diagnostic associative model for teaching the trainee identification of defects and deciding correct product quality (e.g., to decide if the product is acceptable, minor defect, not acceptable). The inductive learning mode is used with the procedural knowledge from the process expert model and expected trainee performance at the rule-based level. Here the trainee needs to reason inductively to understand the appropriate rules (to perform the task of planning a process flow).
  • For the “propose-problem-driven-method” the trainee is asked to solve a problem with practice training units. In Workmanship-MOBAT, this method is appropriate for rote and deductive learning modes. The rote training is used in the same way as commented above. The deductive mode is used with the screen print process simulation and a rule-based expected trainee behaviour. Here the trainee has to reason deductively in order to generate the appropriate rules to set the control panel correctly in different situations.
  • The “propose-test-driven-method” is used to assess a trainee’s ability to make correct product quality decisions. Rote, inductive or deductive learning modes are used by selecting the associative, procedural or principled models for a given goal expertise level and training task.
Figure 6-8 To Present at Different Levels

Figure 6-8 To Present at Different Levels

In summary, the trainer task to present is used for delivering knowledge (to the trainee) below, above or at the same level as shown in Figure 6-8. This task is selecting the generality modelling dimension. A variety of trainer methods are available, supporting different learning modes.

The trainer task primitives to dictate, to coach and to facilitate each implement a training strategy. The discovery strategy does not have a specific task primitive in the didactic tactician. For a discovery strategy all available training unit options are presented, therefore the trainee can freely select his/her own navigation choices.

The tutoring, coaching and facilitating strategies are based on a trainee self-supporting level which may change during a training session. The trainer tasks here provide varying degrees of help and freedom with model navigation for a trainee. One trainer method (called the “propose-focus-method”) is using the scope modelling dimension by selecting a training goal and setting the initial training unit state. Another trainer method (called the “propose-generalise-method”) is using the generality modelling dimension by selecting the appropriate problem solving method.

A third trainer method (called the “propose-simplify-method”) is using the perspicuity modelling dimension by selecting an appropriate case model. In Workmanship-MOBAT, different combinations of these trainer tasks and trainer methods can be used to switch to the appropriate model for a given training situation.

The trainer tasks for MOBAT’s training strategies are described further in Section 7.8.1.

*
Flower Show
  *  
© RuleWorks.co.uk | | Sitemap