Identifying high-level dependence structures using slice-based dependence analysis

Abstract

This thesis presented a framework of the possible combination of approaches for low-level program slicing-based-dependence analysis and high-level concept assignment. Three combination techniques, concept extension, concept abbreviation and concept refinement, are presented and empirical studied to address the problem of program maintenance. The ten C subject programs are studied and more than 600 concept bindings are identified. Dependence based metrics are defined to evaluate three techniques that provides evidence of both advantages and disadvantages.