The method in accordance with the invention is a method making it possible to explain a decision taken by a compensatory multi-criterion aggregation model, according to which a single compensatory multi-criterion aggregation model is established for the application considered, described by a certain number of phenomena and reproducing a determined expert opinion, and it is characterized in that no reference is made in the explanation to the specificities of the aggregation model if the decision taken had been the same on replacing the aggregation model by any other close model, that, if such is not the ease, the aggregation model is simplified by taking into account fewer and fewer phenomena as long as the decision taken remains the same, and that an explanation based solely on the compensation phenomena present in the simplest model previously determined is returned to the expert.

 
Web www.patentalert.com

> Method and apparatus for case-based learning

~ 00357