Knowledge Acquisition in Software Engineering Requires Sharing of Data and Artifacts

Abstract

An important goal of empirical software engineering research is to cumulatively build up knowledge on the basis of our empirical studies, for example, in the form of theories and models (conceptual frameworks). Building useful bodies of knowledge will in general require the combined effort by several research groups over time. To achieve this goal, data, testbeds and artifacts should be shared in the community in an efficient way. There are basically two challenges: (1) How do we encourage researchers to use material provided by others? (2) How do we encourage researchers to make material available to others in an appropriate form? Making material accessible to others may require substantial effort by the creator. How should he or she benefit from such an effort, and how should the likelihood of misuse be reduced to a minimum? At the least, the requester should officially request permission to use the material, credit the original developer with the work involved, and provide feedback on the results of use as well as problems with using the material. There are also issues concerning the protection of data, maintenance of artifacts and collaboration among creators and requestors, etc. A template for a data sharing agreement between the creator and requestor that addresses these issues has been proposed.