Visualizing Software Structure Understandability

Software architecture design is known to be driven by the quality attributes we may want to satisfy. Among them, modifiability plays an important role since software maintenance takes the lion\´s share in the software development costs. However, to successfully maintain a legacy system, the latter must be sufficiently understood so that the maintenance team will not introduce new bugs when correcting others. Then we present a software metric that we called the Autonomy Ratio (AR). We show this dynamic metric to be a good indicator of the system\´s structure understandability. Since we end up with hundreds of values for a single system, we represent these values as a hierarchical map: the "Autonomy Ratio Map". The contribution of the paper is to link the AR metric with theories of software comprehension, to show how the AR Map helps in assessing software structure understand-debility, and to present an empirical validation of it.