Implementation of E x t e n d e d I n d e x e s i n P O S T G R E S Paul M. Aoki ~ Computer Science Division, Department of EECS University of Cafifornia Berkeley, CA 94720-0001 Abstract The vaunted "Spartan simplicity" of the relational model can limit the usefulness of the relational database management system (RDBMS) for non-traditional applications. For example, there is no natural way to model a keyword index for bibliographic informational retrieval (IR), or any other index whose key values are computed from column values, in a standard RDBMS. The extended indexing proposed in [LYNC88a] is intended to support applications that require indexes on computed values. This paper reports on an implementation of this type of indexing using the POSTGRES extensible database management system [STON86a], focusing on two issues: general problems, and the features in POSTGRES that proved helpful in the solution of these problems. and extensions made to POSTGRES beyond those normally supported by the system. Finally, Section 5 summarizes the experience gained from the implementation. 2. Relational Systems for Information Retrieval Information retrieval and database management systems2, as divergent as they often are as research areas,
/lp/association-for-computing-machinery/implementation-of-extended-indexes-in-postgres-emQzowqd7p