Previous: Domain Knowledge Up: Extrapolate from a similar case Next: Modelling operators

Task knowledge

A function structure for ESC at the first task decomposition level is shown in figure .

The first decomposition divides this task into a matching step and a extrapolation step. Match takes a set of requirements (i.e. problem attributes) and produces an applicable theory (i.e. a source case). Extrapolate takes the theory and changes and applies extrapolations based on the difference between the actual requirements and the source requirements.

1. match(+Problem:attribute*, --Source-case:case)

   Returns a source case form the case base that is applicable to Problem. The region of the source problem must cover Problem. Sub tasks are:

   1. find-region-task(+Target-problem:problem, --Region:region). This task takes a problem and searches through the regions by using the region operator.

   2. get-source-case(+Region:region, --Source-case:case). Gets an applicable source case from the case base by applying region operator. It is not clear how overlapping regions are can be dealt with.

2. Extrapolate-task(+Problem:attribute*, +Source-case:case, --Solution:attribute*)

   Extrapolate takes a problem and a matching case and produces an extrapolation of the source solution based on the difference between the problem and the source problem. Sub-tasks are:

   1. determine-difference(+Problem:problem, +Source-case:case, -Differences:difference*).

      Takes a problem and a source case, retrieves the source problem form the case and compares the attribute values. The differences between the attribute values are returned.

   2. find-adjustment(+Difference:difference, -Procedure:)

   3. apply-adjustment.