This paper deals with one problem that needs to be addressed in the emerging field known under the name "computing with perceptions". It is the problem of describing, approximately, a given fuzzy set in natural language. This problem has lately been referred to as the problem of retranslation. Two approaches to dealing with the retranslation problem are discussed in the paper, one based on a pre-defined set of linguistic terms and the associated fuzzy sets, and one based on approximating the given fuzzy set by another fuzzy set that is easier to describe in natural language. The retranslation problem is discussed in terms of two criteria-validity and informativeness.

No abstract