STUDYING THE EVOLUTION OF SOFTWARE THROUGH ENTROPY METRICS

摘要

Based on the information theory definitions and mathematical characterizations of entropy measures, several entropy metrics were suggested for measuring the information content of software. These metrics are the Hartley, Shannon and generalized metrics. In addition to validating the utility of such metrics as indicators of software information content, this paper demonstrates their use in studying the evolution of a large software system. It is shown that studying the evolution of the modules within the system reveals the different module behaviors that are concealed when the full system is viewed at the top level. Thus, it is not enough to consider software evolution from the top level only.