This project aims to combine two state of the art software technologies: genetic algorithms and program slicing. The goal is to use genetic algorithm techniques to construct optimal program slices. However, the project may also consider a wider set of transformations that program slicing and other search-based technologies in addition to genetic algorithms. Although genetic algorithms have been widely explored in the literature and slicing algorithms developed for traditional slicing, there has been little work on the application of genetic approaches to software engineering and no work on applying genetic algorithms to program slice construction.
Program slicing and more general transformation have been applied to a number of problems in software engineering, including, design, testing, re-engineering, maintenance and program comprehension. However, in many applications the choice of transformation strategy is far from obvious. The GASE project aims to view the selection of a `good' transformation strategy to be a search problem to which a genetic algorithm could be applied.
The advantage of the approach lies in the way the genetic algorithm remains invariant, while different metrics and transformation sets can be added and removed as components. By varying the metric, the algorithm will optimise for a different program property. By varying the transformation set, the optimisation strategy will be varied.
The research student appointed to fill this post will be part of the lively and enthusiastic, cross-London research group on Verification and Analysis using Slicing, Testing and Transformation (VASTT).
The project is funded by the Brunel Research Initiative and Enterprise Fund
(BRIEF) for the first
year and by the Department of Information Systems and Computing
(DISC).