Practical Technical Debt Discovery by Matching Patterns in Assessment Graph

Abstract

This article reflects on experiences collected by doing technical debt assessments for many years as a primary job. It argues for a model that represents software source code and other informational artifacts as a graph with metadata describing these artifacts. Technical debt items are discovered with graph matching patterns that represent technical debt discovery patterns. These patterns automate manual work, avoid effort duplication, and boost assessment performance. The overall approach is illustrated with a prototype implementation and a case study.