Using Dynamic Analysis and Clustering for Implementing Services by Reusing Legacy Code

Abstract

Migrating legacy systems towards Service-Oriented Architectures requires the identification of legacy code that is able to implement the new services. This paper proposes an approach combining dynamic analysis and data mining techniques to map legacy code to business processes and to identify code for service implementations based on this mapping. Validating the clustering solution in a first case study resulted in values of 70,6% in precision and 83,5% in recall.