Table II lists a few familiar properties of probability. A one-to-one correspondence can be
seen between the properties in Table II for assertions and probability
and those listed in Table III for questions and bearing. Transformation between
these two ta
Because the issue A of the subject ILU cannot be transmitted without incurring additional
inefficiency or information loss, the best that can be done is for the
source ILU to attempt to optimize itself as a source of information. This can be
achieved if t
Tables II and III summarize this section in terms of some selected properties of
probability and bearing, respectively. Table I forms the basis for deductive computation;
Tables II and III form the computational basis for a computer that operates
inductiv
Perhaps the most important logical property of both questions and assertions
is that of logical implication. If an assertion a implies another b, then if b answers
a question, then a also answers the same question (both will give a true answer
to the subj
Again, the computational goals of the ILU require that its architecture lend itself to the
numerical computation of both b(7|A) and b(X v Y\A). The bearing
b(7|A), or some monotonic function of b(F|A), must be computed to maximize
transmission from the su
The left side of Table I delineates a set of assertive properties that are in oneto-one
correspondence with the interrogative ones on the right. For questions,
remember to interpret the disjunctive operator "v" as common so that, for example, the
common q
One can construct physical examples of each assertive property. For instance,
the property (aAb)vb = b can be interpreted as "if" an observer B is looking for
an assertion b and if a A /? is asserted to B, then B will only observe b, because it
cannot dis
The question A v ~A asks nothing relative to what might be asked by any
observer B, because (A v ^A) v B = Av ^A, whereas A A ~A asks everything
relative to what might be asked by any observer B, because (A A ~A) v B = B.
The term "relative" must be inclu
Regarding inference, the ILU must use the conditional probability p(y\x Aa)
for decision making. The premise consists of the available information given by
the conjunction of jc = xi A JC2 A A jc with a, where a is the basic premise that
delineates precis
As might be surmised, relationships between questions and assertions are logical as well.
As described earlier, any question may be thought of as being defined
by the set of possible answers that may arise through its posing in response to
presented asser