The Evolution
24
zero (thereby assuring that N, by definition, is a non-intersected con-
cept), that P(NI Y) is the highest probability in the Y profile (thereby
assuring that N is the concept most strongly linked to the concept Y),
and that P(X) is sufficiently high to allow the X profile to contain
accurate probabilities (i.e. X has occurred sufficiently often for its
whole-to-whole links to constitute an accurate profile of what is likely
given the occurrence of X).
Now if X is the concept "Florida," Y the concept "eggs," N the concept
"toast," and I the concept "orange juice," then calculating how likely
toast would be in comparison to "orange juice" given the occurrence of
both "Florida" and "eggs" would involve the equation:
P( toast I Florida eggs)_____ _ P(N ∣X) + P(N! Y)_ _____P(N∣ Y)______j
P(orange juice I Florida eggs) P(l∣X) + P(I∣Y) P(I∣X) + P(l∣Y)
which leads to the definition
Type III Ratio _ P(non~intersected concept) _ P(NI Y) .
Pdntersected concept) P(l∣X) + P(l∣Y)
And, as regards a Type II situation, if X is the concept "hospital,"
Y the concept "blade," N the concept "breadknife," and I the concept
"scalpel," then calculating how likely "breadknife" would be in compari-
son to "scalpel" given the occurrence of both "breadknife" and "blade"
would involve the equation:
P(breadknifeI hospital blade) _ P(Y)P(N∣Y)
P(scalpel I hospital blade) P(l∣X)
which leads to the definition
Type II Ratio _ P(non-intersected concept) _ P(Y)P(N∣Y)
Pdntersected concept) P(I∣X)