An initial approach to assessing program comprehensibility using spatial complexity, number of concepts and typographical style

Abstract

Software evolution can result in making a program harder to maintain, as it becomes more difficult to comprehend. This difficulty is related to the way the source code is formatted, the complexity of the code, and the amount of information contained within it. This work presents an initial approach that uses measures of typographical style, spatial complexity and concept assignment to measure these factors, and to model the comprehensibility of an evolving program. The ultimate aim of which is to identify when a program becomes more difficult to comprehend, triggering a corrective action to be taken to prevent this. We present initial findings from applying this approach. These findings show that this approach, through measuring these three factors, can model the change in comprehensibility of an evolving program. Our findings support the well-known claim that programs become more complex as they evolve, explaining this increase in complexity in terms of layout changes, conceptual coherence, spatial relationships between source code elements, and the relationship between these factors. This in turn can then be used to understand how maintenance affects program comprehensibility and to ultimately reduce its burden on software maintenance.