4.1 Control Layer

  The first layer of the RM embodies components which handle incoming presentation goals, presentation commands, and possibly further commands such as start, stop/interrupt, refine goal etc. which allow the user (or a superordinate system) to control the generation process. In that view, control is carried out by an external entity via the Goal Formulation Interface, a component whose task is to convert incoming messages into internal formats understandable by the IMMPS. For example, some systems have a menu-based Goal Formulation Interface which allows the user to choose from a set of preformulated presentation goals, whereas other systems are always triggered by the ``standard goal'' (Present <data>) where <data> is a reference to a certain portion of application data.

External control must not be mixed up with internal control functions which are required to coordinate various generation sub-processes in an implementation of an IMMPS. While early IMMPSs mostly have hierarchical or centralized control structures, there are already implementations which move towards distributed system designs where control is hidden in communication and negotiation patterns of the distributed components. However, the proposed RM intends to capture the structure of a generic generation process. We leave open the architectural organization of system components that one may choose for a concrete implementation.

Regardless of the way in which an IMMPS is actually implemented, there is the inherent need to select from the Goal Formulation Interface the next goal to be achieved, or the next presentation command to be executed. The task of the Goal Selection component (cf. Fig. 4) occurs in multimedia generation for two different reasons: First, it might be the case that the (external) goal formulator poses a set of goals to the presentation system, either all at once or in an undetermined piece-meal fashion. In the latter case, an already started generation process may be interrupted in order to achieve a new incoming goal immediately. The second reason is that presentation goals given as input to the system may be complex, so that they have to be split into sets of less complex goals. While the decomposition of goals will be done by the goal refinement module of the content layer, the control layer still has to decide in which order the subgoals will actually be processed. Deciding the next goal to be achieved can be quite a simple task. It could involve only popping a goal from the discourse model (maintained by the Context Expert, cf. section 5.2). On the other hand, some private knowledge may be present in the Goal Selection component if the decision involves more complex reasoning.

 

Figure 4: The Control Layer