Strongly interacting knowledge granules
Thomas Whalen · 1999
A knowledge granule expressed as a material implication rule is strong to the degree that the fuzzy restriction it imposes is a strict one. When two rules interact strongly, there is no y value that is 100% compatible with both rules for x between the rules' antecedent cores. This paper considers the effects on the behavior of strongly interacting R-implications based on the Schweizer-Sklar parameterized T-norms (1961) as the parameter varies from -/spl infin/ to +/spl infin/. Bases for comparison include (1) the shape of the membership function of the fuzzy set of y values compatible with a given x and the interacting rules, and (2) the defuzzified value of y obtained from this fuzzy set by mean-of-mode and best-kernel defuzzification.