SOFT COMPUTING BASED FRAMEWORK FOR RISK ASSESSMENT IN GLOBAL SOFTWARE ENGINEERING

Abstract

Global Software Engineering (GSE), despite its practical importance, is still an immature field, with palpable shortage of systematic guidelines and best practices in various contexts. This thesis presents an approach which uses Soft Computing paradigm for evaluating risks in GSE. The research work is presented in five parts. First part describes concept evolution, encompassing the introduction of GSE, its importance as an emerging trend, its benefits, challenges and specifically risks involved in adopting GSE, leading to the fruition in the form of Risk Hierarchy. Second part presents the technical drive being followed, in order to assess the categorized risks, which comprises introduction to Fuzzy Logic, Neural Networks, Adaptive Neuro-Fuzzy Inference System (ANFIS), and finally the building blocks, architecture, design and structure of the proposed system emanating from the adoption of above mentioned technologies, to evaluate risks involved in specific GSE project. However, since the system is training based, in connection to that, the process of empirical data collection is discussed along with the mention of training and testing of the system, heading to the results phase, which comprises evaluation of acquired results using error measures and graphical presentation of accuracy of the model. Contingent to an attained accuracy level, the conclusion is drawn that the proposed framework is reliable enough to be adopted by project managers working on globally distributed projects.