Loading…

Collaboration optimization in software process composition

Purpose: The purpose of this paper is to describe an optimization approach to maximize collaboration in software process composition. The research question is: how to compose a process for a specific software development project context aiming to maximize collaboration among team members? The optimi...

Full description

Saved in:
Bibliographic Details
Published in:The Journal of systems and software 2015-05, Vol.103, p.452-466
Main Authors: Magdaleno, Andréa Magalhães, de Oliveira Barros, Marcio, Werner, Cláudia Maria Lima, de Araujo, Renata Mendes, Batista, Carlos Freud Alves
Format: Article
Language:English
Subjects:
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Purpose: The purpose of this paper is to describe an optimization approach to maximize collaboration in software process composition. The research question is: how to compose a process for a specific software development project context aiming to maximize collaboration among team members? The optimization approach uses heuristic search algorithms to navigate the solution space and look for acceptable solutions. Design/methodology/approach: The process composition approach was evaluated through an experimental study conducted in the context of a large oil company in Brazil. The objective was to evaluate the feasibility of composing processes for three software development projects. We have also compared genetic algorithm (GA) and hill climbing (HC) algorithms driving the optimization with a simple random search (RS) in order to determine which would be more effective in addressing the problem. In addition, human specialist point-of-view was explored to verify if the composed processes were in accordance with his/her expectations regarding size, complexity, diversity, and reasonable sequence of components. Findings: The main findings indicate that GA is more effective (best results regarding the fitness function) than HC and RS in the search of solutions for collaboration optimization in software process composition for large instances. However, all algorithms are competitive for small instances and even brute force can be a feasible alternative in such a context. These SBSE results were complemented by the feedback given by specialist, indicating his satisfaction with the correctness, diversity, adherence to the project context, and support to the project manager during the decision making in process composition. Research limitations: This work was evaluated in the context of a single company and used only three project instances. Due to confidentiality restrictions, the data describing these instances could not be disclosed to be used in other research works. The reduced size of the sample prevents generalization for other types of projects or different contexts. Implications: This research is important for practitioners who are facing challenges to handle diversity in software process definition, since it proposes an approach based on context, reuse and process composition. It also contributes to research on collaboration by presenting a collaboration management solution (COMPOOTIM) that includes both an approach to introduce collaboration in organization
ISSN:0164-1212
1873-1228
DOI:10.1016/j.jss.2014.11.036