A Review on Technical Debt Prediction using Static Code Analysis in Open-Source Software Repositories

Abstract

Technical debt describes the long-term cost that software teams take on when they choose quick solutions to save time in the short run. As a codebase grows, this debt accumulates and makes future maintenance harder and more expensive. This paper reviews ten studies on identifying, measuring and managing technical debt, with a focus on static code analysis, a technique that inspects source code without executing it. The reviewed work covers static analysis tools, machine-learning-based detection of self-admitted technical debt, tool comparisons, and prioritisation approaches. The review shows that available tools disagree with each other, use different metrics, and that there is no consensus on how to measure and prioritise debt. Based on these gaps, the paper proposes the direction of a framework that combines several static analysis metrics into one prediction score, validated empirically on open-source software repositories.