Genetic algorithm for database query optimization

Numerous decision support applications have been modeled as set covering and partitioning problems. We propose an extension to the database query language SQL to enable applications of these problems to be stated and solved directly by the database system. This will lead to the benefits of improved data independence, increased productivity and better performance. Six operators, namely cover, mincover, sumcover, partition, minpartition, and sumpartition are extended. In this paper, we presented genetic algorithms for the implementation of access routines for the proposed operators. We found that our genetic algorithm approach for extended operations and query optimization performed well both on the computational effort and the quality of the solutions through a variety of test problems. This approach makes it possible for a DBMS to respond to queries involving the proposed operators in a predicate restricted amount of time.