An architectural framework for expert identification based on social network analysis

Abstract

Social network analysis has been widely used in different application contexts. For example, in Global Software Development (GSD), where multiple developers with diverse skills and knowledge are involved, the use of social networking models helps to understand how these developers collaborate. Finding experts who can help address critical elements or issues in a project is a challenging and critical task. It is especially true in the context of Global Software Development projects, where developers with specific skills and knowledge often need to be identified. In this sense, searching for essential members is a valuable task, as they are fundamental to the evolution of the network. This work proposes an architectural framework for expert identification as a hybrid solution that includes syntactic and semantic analysis in social networks. We seek to address research challenges related to designing recommendation systems when analyzing social structures in the Global Software Development context. In this solution, we define a model for the social network capable of capturing collaboration between developers, incorporate strategies for temporal analysis of the network, explore the network using machine learning algorithms, propose an ontology to enrich the data semantically, and consider a performative approach for high-volume social network analysis methods. We conducted four case studies using data extracted from GitHub to evaluate the proposed approach, as well as a more extensive dataset for the performance studies. The case studies provide evidence that our proposed method can identify specialists, highlighting their expertise and importance to the evolution of the social network.