Construction of BBN Diagrams for Software Process Tailoring

Abstract

Tailoring industrial standards aims to reduce costs and/or improve quality for a particular organization or project. This paper proposes using Bayesian Belief Network (BBN) analysis to support tailoring decision-making under uncertainties. However, there are two major problems associated with the objectivity of BBNs; that is, the construction of the causal inference diagrams and the assignment of probabilities of their dependency relations. We have developed a method to solve the first problem. In general, the relations among different activities, resources, and products addressed in software standards can be expressed more directly in UML diagrams than in BBN’s; such relations include association, aggregation, or inheritance relations. We have developed a scheme to construct BBNs from given UML modeling of software standards for process tailoring purpose. The proposed approach integrating UML and BBNs can also be used to assist decision-making in other software project management activities, such as planning and risk management.