An Efficient Method for Automatic Antipatterns Detection of REST Web Services

Abstract

REST Web Services is a lightweight, maintainable, and scalable service accelerating client application development. The antipatterns of these services are inadequate and counter-productive design solutions. They have caused many qualitative problems in the maintenance and evolution of REST web services. This paper proposes an automated approach toward antipattern detection of the REST web services using Genetic Programming (GP). Three sets of generic, REST-specific and code-level metrics are considered. Twelve types of antipatterns are examined. The results are compared with the manual rule-based approach. The statistical analysis indicates that the proposed method has an average precision and recall scores of 98% (95% CI, 92.8% to 100%) and 82% (95% CI, 79.3% to 84.7%) and effectively detects REST antipatterns.