シナジーのメタファー１

　The fuzzy set of the output value can be cut off at the high point of the membership grade (the output of WHEN-THEN processing rules) (Max/Min method). It is a possibility to transfer the membership grade of the outputs to the individual fuzzy sets of the membership functions of the output value.
The following flow of the membership functions of the ironic distance (arbitrarily selected flow) is given additionally.

IF
μnear (Distance)= 0
μmiddle (Distance) = 0.6
μfar (Distance) = 0.4

So that
μnear (Ironic distance)= 0.4
μmiddle (Ironic distance) = 0.6
μfar (Ironic Distance) = 0

　It appears next to the Max/Min method.
　The sub-areas are outlined to a total area so that we obtain the fuzzy sample area (indicated as the gray area). The concrete value of the distance is determined by applying defuzzification.
Defuzzification is the transformation of a loose fact to a concrete number and value. Generally, one can say that defuzzification provides good results by means of the centroid area.

(59) DEFUZZIFICATION

μ result l ➔
μ result 2 ➔
μ result 3 ➔

　Defuzzification of the distance is applied here, because the distance must be adjusted by the near-far equation. Let’s introduce a defuzzification method called the “mean of maximum” here. This method is suited for rough calculations and the x-coordinate is used as the output value in the middle of the maximum of the sample space.
　We get an ironic distance of 4m.
*Note that an overlapping of sub-areas isn’t applied for this method.

花村嘉英（2005）「計算文学入門−Thomas Mannのイロニーはファジィ推論といえるのか？」より英訳 translated by Yoshihisa Hanamura