Calibrating program slicing metrics for practical use

Program slicing metrics are an important addition to the range of static code measures available to software developers and researchers alike. However, slicing metrics still remain under-utilized due partly to the difficulty in calibrating such metrics for practical use; previous use of slicing metrics reveals a variety of calibration approaches. This paper reports on the effect of including different variables in the calibration (and collection) of slicing metrics. Findings suggest a variety of different results depending on the level of abstraction at which the metrics are collected (system or file), the inclusion or exclusion of ‘printf ’ statements, formal ‘ins’, ‘outs ’ and global variables in the metrics ’ calculation.. 1