SQUAL.AI

Artificial intelligence techniques, particularly machine learning ones, promise to foster breakthroughs in software engineering via the exploitation of large repository data, including code, commits, and human-written comments. This project offers to exploit such techniques to automate quality assurance for highly complex systems, which include configurable systems, AI-enabled systems, or even quantum software. While extremely different, these systems have high variability and uncertainty in common. These two characteristics challenge traditional quality assurance methods, making them inaccurate for smell detection and further complicating the oracle problem in testing. We will combine monitoring such systems and active learning to automatically learn thresholds for smell detection and assess the smells’ impact via automated testing. We will validate our findings on open-source systems fostering open science and reproducibility.
With the support of Wallonie-Bruxelles International.