View source for Talk:Modeling with words

Vilem Novak, Modeling with words

This article gives a short overview over techniques of automated reasoning by means of rules expressed in natural language.

The text is well written, and it is well understandable in nearly all parts. It is in particular quite pleasant to have a reasonable, precise definition of the subject at the very beginning and to see that the text is in full accordance with this concept. (In case of this popular subject, this is far from automatic.)

The web publication is recommended. It is left to the author to take into account the following remarks.

1. FLb is surely well characterized by the four points in the first section. However, I think that FLb will not be the only tool necessary for a "formal theory of human reasoning", but at most one of the ingredients.

Moreover, FTT is a well developped and in the field quite unique theory; still, I assume that in principle, there can be alternatives. This is to say that "the" in "the FLn-mathematical basis ..." should be replaced by "an".

2. After formula (3), "a proper kind of t-norm" might mean "some t-norm". Morever, later, in formula (6), a t-norm is explicitely displayed; for reasons of consistency, the symbol could already be used in (3) as well.

3. In formula (5), the indices X and Y might be missing. Afterwards, the fuzzy sets A and B appear without reference.

4. In the section on Relational inference, "a composition of fuzzy relations" might mean "forming the image under fuzzy relations" or similar.

Below, "forma" should be replaced by "form".

5. The part on Perception-based logical deduction is not as well understandable as the rest. The idea might have been that the text must be kept short. Still, it must be noted that the explanations by means of figures 5--7 are not convincing.

6. Many articles "the"/"a" are missing.