Abstract
A Web service interface is considered as a contract between Web service providers and their subscribers. The subscribers do not have access to the source code of the services but only to the interface containing a set of operations. However, the interface may change over time to meet new requirements. These changes affect the implementation of the subscribers' software. Thus, these clients need to understand the changes introduced to the previous releases of the Web services to co-evolve their own implementation to support the new release. Current studies are limited to the detection of only atomic changes (e.g. add and delete) and not able to detect complex/composite refactorings (merge operations, extract operation, etc.). In this paper, we propose to consider structural and textual similarities, based on a genetic algorithm, when analyzing the evolution of Web services to detect complex changes applied between multiple releases. The validation of our detection technique, on more than 110 releases of 6 real-world Web services, shows an average precision and recall respectively higher than 86% and 89%.