Recommendations for optimized configuration of software product lines

Software product lines provide customized solutions in a more efficient way than individually developed solutions. Based on the product line’s common platform the required elements are selected and parameterized according to the features and goals that are required by the customer. Feature models provide the necessary information for selection and configuration. According to the size of usual product lines and to the high number of possible dependencies, the task of product line configuration is highly complex and error-prone. In this research project, research efforts in three fields are brought together: (1) formalization of requirements, dependencies and environmental constraints, (2) machine learning techniques, and (3) multi-dimensional optimization. They aim at recommendation support for product developers during the configuration of a product line. The support shall include the establishment of proposals for optimized configurations, the resolution of conflicts between competing goals, and checks for conflicts within a configuration and with requirements and environmental constraints.

 

University research project, PhD research project

Duration: June 2017 - Sept 2019

Contact: Yibo Wang, Prof. Matthias Riebisch