Conception of an Automatic Decision Support Platform Based on Cross-Sorting Methods and the Fuzzy Logic for General Use

Abstract

Problems related to the classification of a set of data (methods, tools…), by using the opinions of a committee members, challenge the need for an effective treatment based on the theoretical methods developed through the literature. Indeed, this study was conducted to answer the need identified during the realization of a prioritization study of a graphic modeling methods set in order to develop it into a computing platform. This platform is based on Multi-Criteria Decision Analysis and is destined for practical and general use. The developed platform uses fuzzy pairwise comparisons and cross-sorting methods. It consists of the calculation of two main results: the overall weight of each data to be classified and the consistency index. Regarding the use of fuzzy logic for calculating weights, we opt for the triangular function in the extension of the principle of least-squares logarithmic regression for taking into account the inaccuracy. Finally, the calculation of the normalized weight values can generate an irrational ordering of fuzzy number elements. In our previous work, we have tried to find the conditions on pairwise comparisons values to get rational outcomes, but the calculation was very heavy with the Matlab software. In addition, the modifications to make on the comparisons are multiple, not automatic and depend on the user’s appreciation. The present work gives a classification solution of results in order to keep only input data (comparisons) that respect the rational order of the weights values. Then, an optimization algorithm is programmed to maximize the consistency indexes of this part of results, by allowing a limited margin of modification without greatly altering the initial data.