Computing non-redundant sets of functional dependencies via simplification

Functional dependencies may be considered the skeleton of the Relational Database Model. They describe binary relations between sets of attributes and they characterize the database structure. A way to ensure solid specifications is to transform set of functional dependencies into an equivalent one with less redundancy. In this work we consider a fuzzy extension of the notion of functional dependencies which is consistent with the Armstrong axioms. We improve the well known notion of Duquenne-Guigues basis by enriching the minimality criteria and we introduce a method to remove redundancy in the basis.