Packaging controlled experiments on Software Engineering: an approach based on ontology

Abstract

Experimental Software Engineering (ESE) attempts to evaluate and measure the performance of models and techniques in practical contexts, in order to establish a body of knowledge base to support decision-making. A reliable body of knowledge in Software Engineering requires controlled experiments and their replications to generalize the obtained results. In order to realize collaborative replications between different context groups, some problems arise, like knowledge transfer between these groups. On the other hand, the integration of study results in a common body of knowledge also finds difficulties, because the publications of lab packages differ, they present inconsistent set of information or level of details. There is not an established standard largely used. In this context, it is proposed the use of ontology - the ExperOntology - as a semantic standard to the set of information contained in the package, a standard that can represent the knowledge generated by the experiment. Through the instantiation of several lab packages in the ontology, the endeavor is to evolve ExperOntology's definition to accommodate the lab packages with content. With the support provided by the ontology, it is possible to pursue standardization promoting a better understanding of the information held in lab packages, easing knowledge transfer.