From classification solving problem to constraint solving paradigm (medical computing)

Abstract

Three approaches to the constraint-solving problem in knowledge systems are examined: (1) the development of a domain-dependent constraint solver for qualitative simulation, implementing Kuiper's algorithm in logic programming; (2) the building of a general constraint solver by extending Prolog unification into the inequality and real arithmetic domain; and (3) the use of committed choice nondeterministic languages such as Prolog, CP, and GHC for a concurrent constraint language. Knowledge acquisition for medical diagnostics is considered as an example application.<>