Towards a Framework Based on Open Science Practices for Promoting Reproducibility of Software Engineering Controlled Experiments

Abstract

Experimentation in Software Engineering has increased in the last decades as a way to provide evidence on theories and technologies. In a controlled experiment life cycle, several artifacts are used/reused and even produced. Such artifacts are mostly in the form of data, which should favor the reproducibility of such experiments. In this context, reproducibility can be defined as the ability to reproduce a study. Different benefits, such as methodology and data reuse, can be achieved from this ability. Despite the recognized benefits, several challenges have been faced by researchers regarding the experiments’ reproducibility capability. To overcome them, we understand that Open Science practices, related to provenance, preservation, and curation, might aid in improving such a capability. Therefore, in this paper, we present the proposal for an open science-based Framework to deal with controlled experiment research artifacts towards making such experiments de facto reproducible. To do so, different models associated with open science practices are planned to be integrated into the Framework.