SIGMA - A Tool for Coarse-grained Plan Matching in Automated program Understanding

Plan-based program understanders recognize concepts in the code by comparing the target code chunk against a library of programming plans. For a large plan library, an exhaustive search for a matching plan during concept recognition is not effective. This paper describes a prototype system, SIGMA (the Signature Matching Agent), which supports the comprehension process in the BUG-DOCTOR fault localization tool. SIGMA is a coarse-grained plan matching component. It uses a set of code features called a signature, as an index to retrieve a reduced set of plans from a plan library for matching with the target code chunk. Preliminary experiments with SIGMA show promising results in filtering irrelevant plans from the library. It appears that SIGMA may be an effective pre-processing tool for use prior to attempting the computational intense concept recognition process. We believe that the successful application of the techniques used by SIGMA to plan retrieval could facilitate scaling up automated plan-based understanders for use on large legacy systems.