Coupling artificial intelligence and a system dynamics simulation to optimize quality assurance and testing in software development

Abstract

The allocation of effort to quality assurance and testing is vitally important to the successful development and maintenance of a software system. There is no quantitative method for finding the right allocation policy. The most common methods include allocating fixed percentage of effort for all software projects or using allocations that have been used for similar projects in the past. The benefits of choosing the correct manpower allocation to suit a particular project can be substantial. Using the System Dynamics Model of Software Project Management an optimal quality assurance and testing level for project's development lifecycle can be found. The focus of this thesis is to design an expert system that can be coupled with the model in order to find the optimal allocation of quality assurance and testing effort for a particular project. Two expert system modules were developed, that when coupled with the system dynamics model, will find optimum quality assurance and testing distributions for a software project. The expert system modules were then used to perform sensitivity analysis experiments on the results.