A Refactoring Approach for Too Large Packages Using Community Detection and Dependency-based Impacts

Abstract

Architecture smells are poor design of architectural structures which reveal negative impacts on software quality and increase risks of bugs and implementation problems in the future. Previous research defines and proposes various architecture smells and each one affects software quality in different aspects. Architecture refactoring is a process that removes an architecture smell and improves quality of system without any changes in system behavior. However, removing one of the architecture smells may cause creating of another smell in the system which directly affects software quality. This research proposes a refactoring approach for Too Large Packages smell which is one of the architecture smells by using community detection in extracting process. This process uses for extracting the package which is identified as Too Large Packages smell into small ones and avoids additional impacts from newly appeared dependency-based architecture smells by focusing on Unstable Dependency smell. The proposed approach provides feasible solution to resolve impact from Too Large Packages smell which can be evaluated by comparing number of detected architecture smells and software metrics of before and after refactoring process.