The paper is concerned with an automated tool for software design. It assumes that the user’s requirements have already been documented. The tool identifies program modules in the design of software. Hence it is a tool for “nonprocedural specification of modules, given the logical model of a system.”
The tool, called CAPO (readers will not know that it stands for computer-aided process organization unless they refer to fig. 1), consists of five steps. First, the characteristics of the data flow diagram (DFD) are recorded in six different matrices. In the second step, the weight matrix, which is based on the cohesion of a module, is formed. The next step is concerned with the similarity matrix used to decompose modules into high strength subgraphs. The fourth step is concerned with finding hierarchical clusters. The final step evaluates a given cluster for measuring goodness. A small example is used to compare the solution of CAPO with that obtained manually by Gane and Sarson [1].
The author himself recommends that CAPO must be tested in a large project. In order to apply this technique to a real world problem, industrial process control, CAD/CAM and robotics have to be explored.
With the increase in similar tools (about 23 as reported in Datamation, July 1987), it is not easy to accept any software development tool as a standard. The field of computer-aided Software engineering (CASE) is growing, and a number of such automated tools are expected in the near future. Commercial acceptability of such tools is not clear at this moment.