Extant work on federated databases has concentrated either on schema i
ntegration using semantic models or on query processing using the rela
tional/extended relational model, but both issues have not been addres
sed using a single framework. In this paper, we explore an alternative
approach to database integration using Horn-clause logic as a canonic
al, intermediate representation. First-order logic is used as a langua
ge for expressing the integration of component schemas into a global s
chema, as well as for specifying integrity constraints associated with
global and component schemas. A language based on first-order logic p
rovides the required expressiveness for canonical representation of th
e global schema at the intermediate level. Global to local query mappi
ng is achieved through the procedural interpretation of logic programm
ing. Query optimization is performed through transformations using int
egrity constraints, such as functional, inclusion, and data integratio
n dependencies. PROLOG has been used for implementing some of the algo
rithms described in this paper. The logic-based approach provides a un
iform approach for supporting an integrated view of a set of heterogen
eous databases, as well as its implementation.