Characterization of foods by inexact rules: The SEXIA expert system

SEXIA is an expert system that uses a new methodological approach to identify foods, particularly olive oils according to varieties, olive zones and denominations of origin. The methodological approach provides identification tools, associating a confidence degree or a belief interval to the final hypotheses. The certainty factor and the Dempster–Shaffer theory, with some modifications, have been implemented in SEXIA. The computer can work with 50 chemical parameters whose data have previously been acquired by the food analyst via a dialogue in the Spanish language. The system has been verified with 144 olive oil samples. In this paper some results obtained for distinguishing the Arbequina variety from other varieties using SEXIA and the BMDP stepwise discriminant analysis program are presented. Finally, promising directions for future research are suggested.