e-VOL BPMN: uma técnica para auxiliar a evolução de modelos e a aprendizagem da notação BPMN

Abstract

Software maintenance is an important operation that consumes a large part of the resources involved in the software life cycle. In this context, software models have a fundamental role in maintenance because they help to understand the software before performing maintenance. Software maintenance includes fault correction or addition of new functionalities. Among the software models, business process models are used to support the understanding of software-related processes. A standard notation to describe these models is the Business Process Model and Notation (BPMN). It is common to find, in the software industry, outdated models compared to the current software version. This is because usually only the source code is updated during maintenance. Outdated software models are especially critical in evolutionary maintenance, where new features are added and the previous model becomes outdated. We performed a systematic mapping in order to verify whether there are technologies proposed to facilitate the maintenance of BPMN models and, in this way, to contribute to update the process models during the maintenance. The objective of the systematic mapping was to find experimental evidences of the maintenance of BPMN models in the literature. The results revealed that little is being done to support the maintenance of BPMN models, despite the importance of keeping models updated. In this context, we developed a technique to support the evolutionary maintenance of BPMN models, called e-VOL BPMN. The purpose of e-VOL BPMN is to assist professionals (engineers and software developers) with low BPMN knowledge in the evolution of BPMN models. We performed an experimental study to evaluate the e-VOL BPMN technique with an experimental group and a control group. We evaluated the models of the experimental study through semantic completeness, which is a metric that shows whether the relevant aspects of the software are specified in the model. The results revealed that the experimental group developed more complete models than the control group. The experiment showed indications that the e-VOL BPMN technique assists in the evolution of BPMN models. Based on the results of the experimental study, we improved our technique, developing the second version of e-VOL BPMN, which we also evaluated experimentally.