| Abstract: | The lack of an existing file design methodology for data base systems is established. A file design procedure for a class of transaction oriented data base systems is then developed. An enumerative algorithm is described. This algorithm involves rearrangement of a data element-query set matrix to locate clusters of data elements, thereby significantly reducing the number of possible file designs without eliminating optimal designs. The reduced solution space is then systematically searched for the optimal design. Additionally, a programmed solution of the algorithm is presented to demonstrate the utility of the concept. Several sample data bases are described and analyzed. |
|---|