Architecture of FAME: fuzzy address matching environment

There have been a considerable number of developments and applications of fuzzy set theory in the database area. In this paper the use of databases with imprecise information for matching and correcting of addressing is described. The context for the motivating application is that of correcting ordinary postal addresses. However other correction applications such as independent shippers, phone numbers from inexact information etc. are also feasible. The basic components of an appropriate application are that it is possible to understand the structure of an imprecise input and then to be able to match such an input to a set of highly structured patterns which themselves have some uncertainties. The paper looks at the basic architecture of a fuzzy information system prototype for address matching and discusses the main approaches to components of the architecture