Forming the optimal union of multiple databases

摘要

Many organizations, best illustrated by libraries, have access to hundreds of databases possessing varying degrees of complimentary and overlapping records. While users may receive positive marginal benefits from data duplications, all databases and their services cannot be supported because of resource limitations. This paper attempts to conceptualize the problem of database collection in an environment of multiple databases, with differential content and performance characteristics, diverse users, and limited resources. It defines the database collection problem as a constrained zero-one integer programming problem and solves for the optimal combination or union of databases. Several extensions are shown, where special conditions are imposed on the relationships between databases and/or their availability.