Sistema especialista no apoio à classificação de criticidade de versão de software

Abstract

The correct execution of a software release management process can be accomplished by applying methodologies and good practices such as Business Process Management (BPM), Six Sigma, Information Technology Infrastructure Library (ITIL) and Project Management Body of Knowledge (PMBOK). There is a need on the part of human process performers to rate the criticality of each software release. However, the subjectivity of this classification may be present in function of the experience acquired by the specialists over the years. In order to reduce subjectivity in the process, the Expert System (ES) can be applied, which stores the knowledge of human specialists. The objective of this work was to reduce subjectivity in software version criticality rating with the support of Expert System. The research was divided into five phases: Phases 1, 2, 3 and 4 contemplated the documentary research of the application in the process of the methodologies and good practices mentioned above. Results are now measured using Key-Performance Indicators (KPIs). However, the correct execution of the process showed that the software release packages present differences in their complexity, and the classification of these versions by human specialists presented subjectivity. Phase 5 dealt with the development and application of ES in the process for reducing subjectivity. A questionnaire was elaborated with the objective of obtaining the criticality opinions classified in High, Medium and Low software version of each expert to assist in the elaboration of the ES production rules. To validate the ES, a process homologation database was used, with results obtained that met the criticality criteria previously known only to specialists, reducing subjectivity in the classification. With the validated ES, the ES was applied to a production database, yielding results that continued to meet the criticality criteria known to the experts. Then, a questionnaire was applied to the specialists in order to obtain the final perception of satisfaction with the use of ES as well as the results of criticality rating obtained, presenting feedback with positive results. It was concluded, then, that ES supported the reduction of subjectivity in the software version criticality rating.