A Human-Centric Quality Assurance Process for Open Source Softwa...
Talkat Open Source Summit Japan + AI_dev: Open Source GenAI & ML Summit + Automotive Linux Summit 2025 · 2025-12-08..12-10·
summary
An automated defect-prediction engine augments traditional size/complexity metrics with memory decay and alertness/time-of-day signals computed from git timestamps and time zones, scoring likely-defective functions for targeted review. Applied to 21 high‑criticality repositories (examples: NASA Trick and Condor, Linux/LLVM in the study), the method surfaces context-dependent issues—copied comments that corrupt generated docs, dead stores missed by static tools—and the paper reports 48% more defects found versus classic metrics; workflow plugs into pre-merge triage and continuous improvement.
AI-written summary · about
chapters (8)
- 0:00Human-centric QA in open source
- 6:26Human factors in software quality
- 12:59Defect prediction with cognitive metrics
- 20:25Three-phase lightweight QA process
- 22:13Elusive defects found
- 28:14Why traditional methods fail
- 30:47Integrating QA into pre-merge
- 34:46Key takeaways
related talks
36mThe Code Quality Advantage: How Empirical Data Shatters the Speed vs Quality Myth
25mBad Behaviour: Automating System-level Safety Testing Using Open Source Tools
01h 21mTradeoffs in the Software Workflow
40mReal Quality Assurance from an Agile and Lean Viewpoint
19mHow can we improve quality on a project?
59mCode Reviews - Why, what and how
01h 19mPNSQC 2013 - Capers Jones
46mQuality Engineering for DevOps Teams