Database mining for the discovery of extended functional dependencies

The paper discusses the semantics of some basic extensions of the concept of a functional dependency when fuzzy sets are used. Extended FDs are especially studied from a knowledge discovery point of view. Two types of extensions are considered. The first one is based on a relaxation of equality into resemblance. The second one is based on a modification of the granularity of the domains and the use of gradual rules. A mining algorithm suited to the discovery of extended FDs of the latter sort is outlined, and the interest of the rules discovered is then questioned. It is shown that, in some cases, additional information can be extracted from the obtained set of rules