ARISE: A framework for automated risk identification in Scrum using memory-based recommendation

Context: Although Scrum relies on empiricism to manage risks, it lacks mechanisms for organizational learning across multiple teams, often confining risk knowledge to individual projects. Objective: This article introduces ARISE (Automated Risk Identification in Scrum Environments), a framework that captures and reuses risk knowledge using a memory-based recommender system to suggest relevant risks from historical data. Methods: Developed in collaboration with an industrial partner, ARISE was evaluated in two phases: (1) an offline experiment with 17 mobile projects exploring similarity metrics and neighborhood sizes under a k -NN model, using precision, recall, and F 1 -score; and (2) a TAM-based evaluation with 24 practitioners to assess perceived usefulness and ease of use. Results: The configuration using Manhattan distance with k = 2 performed best, achieving 63% precision and 80% recall. This small neighborhood size proved optimal because risk profiles in the studied projects are highly specific; a larger k introduced “noise” from less similar projects, whereas k = 2 captured the most relevant local patterns. Conclusion: ARISE bridges the gap between adaptive risk handling and systematic knowledge management, enhancing proactive identification while preserving Scrum values. Editor’s note: Open Science material was validated by the Journal of Systems and Software Open Science Board .