Personalized Web Services Interface Design Using Interactive Computational Search

Abstract

Most of successful Web services evolve through a process of continuous change due to several reasons such as improving the quality, fixing bugs and adding new features. However, this evolution process may weaken the design of the Web service’s interface by including a large number of non-cohesive operations and make it unnecessarily complex for users to find relevant operations to be used by their services-based systems. In this thesis, we propose a remodularization recommendation approach that dynamically adapts and interactively suggests a possible modularization of the Web services interface design to users/developers and takes their feedback into consideration. Our approach uses an interactive multi-criteria decision making algorithm, based on interactive NSGA-II, to find a set of good design interface modularization solutions that find a trade-off between improving several interface design quality metrics (e.g. coupling, cohesion, number of portTypes and number of antipatterns), maximizing the reuse of user-interface interaction history patterns identified from previous releases and satisfying the interaction constraints learnt from the user feedback during the execution of the algorithm while minimizing the deviation from the initial design. We evaluated our approach on a set of 22 real-world Web services, provided by Amazon and Yahoo. Statistical analysis of our experiments shows that our dynamic interactive Web services interface modularization approach performed significantly better than the state-of-the-art modularization techniques.