AN EFFICIENT APPROACH TO IDENTIFY KEY CLASSES OF SOFTWARE TO ASSIST INITIAL PROGRAM COMPREHENSION

Abstract

The constant modification of software systems, the increasing size of the software and the expensive development process are the factors that are responsible for the increase in the amount of the effort that is being expended on the maintenance phase. Mostly each maintenance cycle is performed to achieve a specific goal, for instance improving the efficiency of a procedure, provision of new application features, assembling existing components into the new software and so on. A programmer will not be able to achieve any of the above goals unless he understands the particular software at a sufficient level of detail that allows him to implement the desired change in the system. The process of exploring the software and its associated artifacts with the aim to gain knowledge about the inner workings of the system for carrying out the necessary changes in the system is termed as program comprehension. The process of building an understanding of the existing system is time consuming and takes around 40% of the allocated time for a maintenance task. How a new software developer proceeds to build an initial acquaintance with the original software, differs a lot and depends on a number of factors like the experience of the individual, the size and type of the software, the level of detail required to modify the system and so on. Numerous efforts have been made to reduce the time consumed in the program comprehension process by providing support to the programmer during this phase. The key contribution of this thesis is a heuristic approach that can aid the programmer in searching the key classes of software that are ideal nominees for the initial stages of the program comprehension process. Full automation of the process of developing an understanding of the program is not possible since it involves human learning activity. Therefore, it has been suggested that the specialized tools should help the programmer discover the software amicably. The segments of the program that can be attractive from the comprehension viewpoint must be brought into the notice of the programmer by the exploration tools. In our case the program discovery tool should pinpoint the key classes of the object-oriented system that are fundamental to its design.