A New Approach for Software Quality Assessment Based on Automated Code Anomalies Detection

Methods and tools to support quality assessment and code anomaly detection are crucial to enable software evolution and maintenance. In this work, we aim to detect an increase or decrease in code anomalies leveraging on the concept of microstructures, which are relationships between entities in the code. We introduce a tools pipeline, called Cadartis, which uses an innovative immune-inspired approach for code anomaly detection, tailored to the organization's needs. This approach has been evaluated on 3882 versions of fifteen open-source projects belonging to three different organizations and the results confirm that the approach can be applied to recognize a decrease or increase of code anomalies (anomalous status). The tools pipeline has been designed to automatically learn patterns of microstructures from previous versions of existing systems belonging to the same organization, to build a personalized quality profiler based on its codebase. This work represents a first step towards new perspectives in the field of software quality assessment and it could be integrated into continuous integration pipelines to profile software quality during the development process.