LEARNING EVENTS FROM SIMULATION MODEL RUNS: FROM PROCEDURAL MODELSTO RELATIONAL ABSTRACTIONS
Süleyman Sevinç · International Journal of General Systems · 1991
Simulation models are largely procedural and are proposed to be used to coordinate the-actions of an autonomous system. The problem, however, is the fact that procedural models lack a well-developed reasoning system. This paper suggests that relational representations, which have a relatively well-developed logical basis for deductive reasoning, be extracted from simulation models and be used as metaknowledge to reason about aspects of simulated behavior generated by procedural model simulation runs.