Two sides of fuzzy databases: flexible queries and imprecise information management

Abstract

Very often, the generic term "fuzzy database" is meant for two different purposes. One is the idea of being provided with queries inside which preferences are used instead of classical Boolean conditions. Basically, such queries are addressed to regular databases and any element of the result is associated with a d e m e of satisfaction used to rank-order the answers. On the other hand, one may face a situation where some pieces of information are not known precisely (automated recognition of objects in images, data fusion, linguistic descriptions, ..., etc). This leads to store and query a new type of data with respect to that dealt with in commercial systems. When fuzzy sets (indeed possibility dis~butions) are used to model such data, one gets fuzzy databases. Obviously, flexible queries can also be used against fuzzy databases, but the two issues are fundamentally independent.