A soft multi-criteria decision-making approach to assessing the goodness of typical reasoning systems based on empirical data

Abstract

A soft multi-criteria decision-making system for the assessment of typical decision algorithms applying empiric data was constructed. The evaluation criteria were simplicity, correspondence with human reasoning, correspondence with reality, content validity and residual distribution. By virtue of our soft computing (SC) approach, we were able to computerize the assessments. The decision-making system was used to examine the Zimmermann–Zysno data of tiles by applying both conventional and those based on SC decision algorithms. The conventional algorithms comprised linear and non-linear regression analysis, whereas a fuzzy and a neuro-fuzzy system represented the SC approach. The SC approach seemed to yield better results within the bounds of the given criteria.