What is it about?
Next release problem and feature selection is a critical challenge in continuous software deployment. In this research we propose a scalable feature prioritization framework that leverages rules derived from feature perplexity, probability, and customer satisfaction.
Featured Image
Photo by Jona on Unsplash
Why is it important?
A scalable feature prioritization framework that leverages rules derived from feature perplexity, probabilistic relevance, and customer-oriented criteria. Our experiments show that we maintained linear complexity (O(m)) while compared to cubic complexity of traditional graph based models (O(m^3)). We have 85.56% of accuracy with superior decision quality (11.28%) of false negatives.
Perspectives
This research presents a proof on how rule based reinforcement learning will help in designing and scaling tasks with high complexity (O(n)). Particularly, when the task has to do with selection of new innovative features balancing the customer needs and companies goals, this method shows to be achieving a really good accuracy with less complexity.
Hemanth Gudaparthi
Governors State University
Read the Original
This page is a summary of: A Scalable Rule-Based Deep Reinforcement Learning Framework for the Next Release Problem, October 2026, ACM (Association for Computing Machinery),
DOI: 10.1145/3832783.3834348.
You can read the full text:
Contributors
The following have contributed to this page
