Constructing information systems based on schema reuse

Information systems, which are based on database management systems, consist of two major parts: A alztabase schema and applications. We extend the view of sofmare reuse to encompass not only application programs, but also the database schema. In ~his pape~ we develop a technique for clustering and retrieving schema components based on their semantics. Our technique is based on both faceted classl@cation and neural network pattern recognition. The advantages of our approach include the user only needs to index a small portion of the components using faceted classification yet all other components are also available for retrieval by automated clustering. An application using this technique in building a data warehouse is discussed.