Search based optimization of requirements interaction management

Abstract

There has been much recent interest in Search Based Optimization for Requirements Selection from the SBSE community, demonstrating how multi-objective techniques can effectively balance the competing cost and value objectives inherent in requirements selection. This problem is known as release planning (aka the 'next release problem). However, little previous work has considered the problem of Requirement Interaction Management (RIM) in the solution space. Because of RIM, there are many subtle relationships between requirements, which make the problem more complex than an unconstrained feature subset selection problem. This paper introduces and evaluates archive-based multi-objective evolutionary algorithm, based on NSGA-II, which is capable of maintaining solution quality and diversity, while respecting the constraints imposed by RIM. © 2010 IEEE.