Re: "Most appropriate terms"

Peter Elsea (elsea@CATS.ucsc.edu)
18 May 1995 17:48:15 GMT

In article <3pc5sq$lec@kralle.zdv.Uni-Mainz.DE> Markus Beckmann,
beckmann@Informatik.Mathematik.Uni-Mainz.DE writes:
>My question is are there any (fuzzy) strategies for
>assigning such terms to objects in the universe?
>
>We might say that a term t is most appropriate for an
>object u iff
>
> t(u) = max_{t \in T} t(u)
>
>but that lead to a "crisp" relation like
>
> "u is called 'cold'" if u \in [-30!,10!]
> "u is called 'warm'" if u \in [10!,25!]
> "u is called 'hot'" if u \in [25!,50!] .
>
>
>Any suggestions?

Your statement t(u) = max_{t \in T} t(u) is a fuzzy tautology.

The point to describing cold, hot, and warm in fuzzy terms is that you
can do further operations. For instance, call the set you derived
{ucold,uwarm,uhot} 'whatsitlike'. I could generate a whatsitlike set for
each of six different cities in California, and then use min-max or add
them up to get a set of whatsitlike in California. Do the same for
Germany. Then taking the max position in each allows us to say with some
authority, It is warm in California and cold in Germany.

Peter Elsea
UCSC

------------------------------