We suppose that the solution of a particular problem needs a particular context, whose components may be more or less necessary. This context can be computed on the image or on the G.I.S, and will be finally considered as satisfied with a global realization degree.
At the beginning of the process, all points (x,y) are
assigned an identical "Solution Degree" DSINIT(x,y)
for the problem (value 0. : indifferent). Then, applying successively all
knowledge units (production rules), the system updates
these degrees with real values belonging to the interval [-1.,+1.]
.
-1. | The point is absolutely not a solution. |
0. | The point is indifferent for the problem solution. |
+1. | The point is the best possible solution. |