Adoption of Software Product Lines for Managing Experiment Replications in a Family of Experiments

Abstract:

ContextExperimental software engineering experiments need to be replicated to assure that the acquired knowledge is mature enough for later use. No formal protocol for the reuse of part or all of the experimental products is followed in the execution of replications, making it troublesome to plan other replications or gather information on completed replications. ObjectiveThe aim of this paper is to adopt software product lines to manage the experiment replications for application and validation in a code evaluation technique comparison experiment and a personality quasi-experiment. MethodThe study was conducted within a research group with lengthy experience in experiment replication. This group administers some experiments with replication management problems. This context is suitable for applying action research leading to the development of a framework that provides techniques, symbology and models for developing the experiment product line. ResultsWe developed an experiment and quasi-experiment line. We calculated the workload in person-hours required for their generation. Finally, we found that the experiment product line was useful for managing the quasiexperiment replications. ConclusionThe framework was useful for developing the experiment product line. The experiment product line is composed of models identifying the status and traceability of the replications, their common and variable materials, and the valid configurations of the replications of an experiment. This information is useful for reusing the products to execute new replications in different contexts.

Año de publicación:

Keywords:

Fuente:

google