Web Service API Anti-patterns Detection as a Multi-label Learning Problem

Abstract

Anti-patterns are symptoms of poor design and implementation solutions applied by developers during the development of their software systems. Recent studies have identified a variety of Web service anti-patterns and defined them as sub-optimal solutions that result from bad design choices, time pressure, or lack of developers experience. The existence of anti-patterns often leads to software systems that are hard to understand, reuse, and discover in practice. Indeed, it has been shown that service designers and developers tend to pay little attention to their service interfaces design. Web service antipatterns detection is a non-trivial and error-prone task as different anti-pattern types typically have interleaving symptoms that can be subjectively interpreted and hence detected in different ways. In this paper, we introduce an automated approach that learns from a set of interleaving Web service design symptoms that characterize the existence of anti-pattern instances in a service-based system. We build a multi-label learning model to detect 8 common types of Web service anti-patterns. We use the ensemble classifier chain (ECC) model that transforms multi-label problems into several single-label problems which are solved using genetic programming (GP) to find the optimal detection rules for each anti-pattern type. To evaluate the performance of our approach, we conducted an empirical study on a benchmark of 815 Web services. The statistical tests of our results show that our approach can detect the eight Web service antipattern types with an average F-measure of 93% achieving a better performance compared to different state-of-the-art techniques. Furthermore, we found that the most influential factors that best characterize Web service anti-patterns include the number of declared operations, the number of port types, and the number of simple and complex types in service interfaces.