The proposed representation for air campaign plan objectives is basically an extension of the present representation used in ACPT. The main goals of the proposed representation:
ACPT currently distinguishes the following components of an objective:
The representation proposed in this document extends the above representation in order to address two main problems. First, it proposes a more detailed and structured representation for the Title field, that also makes superfluous the fields Effect and Action. Second, it proposes a new field Sequence restrictions, that clears up a difference between prioritization and sequencing which is currently not made and which (according to CHECKMATE) has caused considerable confusion among ACPT users.
The Title field actually contains a statement or specification of the objective -- for instance, Disrupt SAM sites in the Western region. This is presently a free text field in ACPT. The action field and effect fields hint at specifying the desired effect and action types. However, they fall short of specifying the objective in adequate richness of detail. Further, they are filled from a menu of options which is currently extremely minimal and consequently poor.
The proposed representation provides structure to what is currently the title, decomposing it as follows:
This structure is inspired in case grammars. The basic idea of case grammars is that there is normally a limited number of roles (called thematic or case roles) that an argument of a verb can play with relation to the verb. While case grammars have been dismissed as a general solution for natural language interpretation, they can be an interesting and powerful device in restricted settings such as the one we have in the air campaign planning domain.
The proposed structured representation eliminates the fields action, effect and location of the current ACPT representation.
ACPT allows the specification of the relative priorities of objectives, which are inherited by their children. If Air Objective O1 has priority over Air Objective O2, the Air Tasks which are specified as their children will also obey that order, i.e. all tasks children of O1 have a higher priority over all tasks children of O2. The priorities may be overridden when the level of targets is reached, i.e. the user may reprioritize any target at his/her will.
While this mechanism has been useful, interviews with CHECKMATE have revealed that users often confuse priority with sequence. In general, the fact that objective O1 has a higher priority than O2 does not mean that it should be executed first. Indeed, if O2 is an objective whose effect is a prerequisite for executing O1, O2 is the one that should be executed first.
We believe that one of the reasons why this confusion exists is
because users are currently not allowed to specify that an objective
is to be accomplished before (or after, or at the same time than)
another. Therefore, we propose to introduce a new field Sequence
restrictions, where the user can specify (if known) which objectives
are to be accomplished before/after/simultaneous with
others.
Further, we have noticed that it is frequently difficult for air campaign planners to make definite sequencing commitments. These commitments have some gradation: sometimes one objective must be accomplished before another is to be pursued, while in other cases it is better but not strictly obligatory to do so. For this reason, we have specified that the three basic sequence relations (before, after and simultaneous with can be modified by one of the following: must, better, better not, must not. These modifiers specify the ``hardness'' or ``certainty'' of the sequence restriction, i.e. how serious they should be taken in the actual sequencing of the objectives and activities later on.