O uso de crowdsourcing em revisões sistemáticas da literatura na engenharia de software

Abstract

Systematic Literature Reviews (SLRs) have been adopted in Software Engineering (SE) to provide a summary of evidence. However, conducting an SLR remains time-consuming and expensive, especially with regard to the hours dedicated by researchers. It is necessary to explore faster and more efficient approaches to support SLR conduction. One approach is Crowdsourcing, since it divides tasks between a large number of researchers. Objective: To identify and summarize the body of knowledge on crowdsourcing to support the conduction of SLRs in SE. Method: Initially, a Systematic Mapping (SM) was carried out, identifying studies involving crowdsourcing in SLRs in SE. In the sequence, a survey was conducted with 39 SE researchers to identify their perception when performing SLR collaboratively. Results: Speed up the SLR process; reduce bias through broad participation, and expand team's experience were the most potential benefits associated with the use of crowdsourcing in SLR. The main challenges are associated with quality control of results. Conclusions: Despite the challenges, crowdsourcing can be successfully used in SLR context. More empirical research is needed on how to use crowdsourcing to conduct SLR in SE and how to minimize the identified challenges.