=ADD=
=reftype=
14
=number=
99-13
=url=
ftp://ftp.risc.uni-linz.ac.at/pub/techreports/1999/99-13.ps.gz
=year=
1999
=month=
05
=author=
Szigeti; Ferenc + De Sarrazin; Gisela + Cardillo; Juan
=title=
Computer aided problem solving via optimization
=abstract=
Assuming a finite  set of tools with a rule
based implementation, we define a language. The active tools are
defined at each word of language. Suppose that a valuation
function is given on our language, which indicates whether a
sequence of tools (word) solves a certain task. Then we construct
a minimal realization, a graded state space representation of the
valuation function. Finally, by calculating an ``optimal control"
problem for a multi-(two) objective function, we can compute that
after making a ``move" (applying a new active tool) the solution
of our task is approaching, moving away, or that the step does not
affect our ability to solve our task. Hence, after the
implementation of our algorithm, the problem solving is supervised
by an intelligent computer, capable of the evaluation and decision
making.
=keywords=
language, discrete events dynamics system, minimum principle, multi-objetive function, intelligent dialogue.