A hybrid ant colony tabu search algorithm for solving next release problems

Abstract

Next Release Problem (NRP) is a challenge in software engineering to define which set of requirements are to be developed in the next release of a software product taking in consideration several constraints such as the cost of development, user’s significance, and constraints related to scheduling, dependencies between requirements and available expertise. Solving this problem will help software engineers to make decisions on the set of requirements to include as features in the next release of a software product. This paper proposes a hybrid of Ant Colony Optimization (ACO) algorithm and Tabu Search (TS) for solving NRP using a cost-value model for requirements. A fitness function with two objectives was considered to maximize users’ satisfaction and to minimize the cost of developing the requirements requested by users. The hybrid Ant Colony Optimization Tabu Search (ACOTS) algorithm is based on Ant Colony Optimization (ACO) algorithm while it employs the history keeping strategy of Tabu Search (TS) when constructing new solutions (local search spaces) for each initial solution generated by each ant. The procedure of the hybrid algorithm starts by generating random solutions that serve as a pivot for all ants of the colony which is based on the pheromone information, the set objectives in the fitness function and problem specific local heuristic information associated with each of the objectives. The output of the hybrid ACOTS is a set of promising optimal values which are the total number of the set of requirements from which a subset is to be selected. The results of the experiments showed that the application of ACOTS yielded larger and better sets of results than existing methods (ACS, Ant System and Tabu Search). The application of ACOTS also enables an easier parameter tuning (budget, number of requirements).