Improving Web Services Design Quality Using Heuristic Search and Machine Learning

Web services evolve over time to fix bugs or update and add new features. However, the design of the Web service's interface may become more complex when aggregating many unrelated operations in terms of context and functionalities. A possible solution is to refactor the Web services interface into different modules that help the user quickly identifying relevant operations. The most challenging issue when refactoring a Web services interface is the high number of possible modularization solutions. The evaluation of these solutions is subjective and difficult to quantify. This paper introduces the use of a neural network-based evaluation function for the problem of Web services interface modularization. The users evaluate manually the suggested modularization solutions by a Genetic Algorithm (GA) for a number of iterations then an Artificial Neural Network (ANN) uses these training examples to evaluate the proposed Web services design changes for the remaining iterations. We evaluated the efficiency of our approach using a benchmark of 82 Web services from different domains and compared the performance of our technique with several existing Web services modularization studies in terms of generating well-designed Web services interface for users.