Hybrid Intelligence in Software Release Planning

  • Authors:

  • Günther Ruhe;An Ngo

  • Affiliations:

  • University of Calgary, 2500 University Drive NW, Calgary, AB T2N 1N4, Canada. ruhe@ucalgary.ca, ango@cpsc.ucalgary.ca;The University of Calgary, 2500 University Drive NW, Calgary, AB T2N 1N4, Canada. ruhe@ucalgary.ca, ango@cpsc.ucalgary.ca

  • Venue:
  • International Journal of Hybrid Intelligent Systems
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

There is a growing recognition that an incremental approach to software development is often more suitable and less risky than the traditional waterfall approach. Delivering software in an incremental fashion suggests better customer satisfaction and reduces many of the risks associated with delivering large software projects. In this paper, we consider the problem of deciding which requirements should be assigned to which release. The proposed hybrid approach called EVOLVE* improves existing methods for release planning by combining the strength of mathematical models with the subtleness of experts' knowledge and judgment. It makes use of different computationally intelligent techniques such as evolutionary computing and principles of multi-criteria decision aid. This is combined with appropriate involvement of human intelligence. EVOLVE* consists of three main phases called modeling, exploration, and consolidation. Different from former algorithms of the EVOLVE family, our new approach plans only two releases in advance, i.e., each requirement is assigned to one of the following three categories: "next release", "next but one release", "not yet assigned". EVOLVE* aims to achieve maximum stakeholder satisfaction. Our iterative procedure allows intelligent search of most promising solutions under the competing criteria of time, benefit and quality as described by the "magic triangle". The complete approach is illustrated by a case study example.