Automatic Identification of Self-Admitted Technical Debt from Different Sources
Technical debt refers to taking shortcuts to achieve short-term goals while sacrificing the maintainability and evolvability of software systems. Nowadays, there is a trend that researchers focus on technical debt that is explicitly admitted by developers, namely Self-Admitted Technical Debt or SATD. However, there are no approaches available for automatically identifying SATD from multiple sources. Therefore, we propose and evaluate an approach MT-Text-CNN for SATD identification in multiple sources. Our findings show that our approach outperforms baseline approaches which achieves an average F1- score of 0.611 when detecting four types of SATD (i.e., code/design debt, requirement debt, documentation debt, and test debt) from source code comments, commit messages, pull requests, and issue tracking systems.
READ FULL TEXT