Estimation of certainty factor of knowledge with fuzzy Petri nets

Presents a method for automated estimation of certainty factors of knowledge from the proven and historical databases of a typical reasoning system. Certainty factors, here, have been modeled by weights in a special type of recurrent fuzzy neural Petri net. The beliefs of the propositions, collected from the historical databases are mapped at places of a fuzzy Petri net and the weights of directed arcs from transitions to places are updated synchronously following the Hebbian Learning principle until an equilibrium condition is reached. The model for weight adaptation has been chosen for maintaining consistency among the initial beliefs of the propositions and thus the derived steady-state weights represent a more accurate measure of certainty factors than those assigned by a human expert