Mining the Maintenance History of a Legacy Software System

  • Authors:
  • Jelber Sayyad Shirabad;Timothy C. Lethbridge;Stan Matwin

  • Affiliations:
  • -;-;-

  • Venue:
  • ICSM '03 Proceedings of the International Conference on Software Maintenance
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

A considerable amount of system maintenanceexperience can be found in bug tracking and source codeconfiguration management systems. Data mining andmachine learning techniques allow one to extract modelsfrom past experience that can be used in futurepredictions. By mining the software change record, onecan therefore generate models that can be used in futuremaintenance activities. In this paper we present anexample of such a model that represents a relationbetween pairs of files and show how it can be extractedfrom the software update records of a real world legacysystem. We show how different sources of data can beused to extract sets of features useful in describing thismodel, as well as how results are affected by thesedifferent feature sets and their combinations. Our bestresults were obtained from text-based features, i.e. thoseextracted from words in the problem reports as opposedto syntactic structures in the source code.