Machine learning techniques to model the impact of relational communication on distributed team effectiveness

Purdue University, West Lafayette IN

Investigators

Abstract

Social scientists are anxious to understand the relational aspects of distributed teams, but examination of interpersonal communication in such settings has been limited by the statistical techniques available for analysis. This project exploits recent advances in the field of machine learning to study relational communication flow in distributed or virtual groups, analyze the impact of communicationon effectiveness, and, as a result, to advance understanding of the complex interdependencies among virtual-group members. The PIs posit that it is the dependencies among team members that hold the key to understanding the processes that impact the success (or failure) of distributed teams. The project involves a two-step approach to theory formation and refinement, combining large-scale observational data and experimental studies. The research team will first extract and analyze publicly available data from open-source software development projects to develop models of effectiveness based on relational patterns of communication among members. They will use the results of this analysis to develop targeted hypotheses for empirical evaluation and conduct laboratory experiments testing these. The results of this project should provide a more comprehensive understanding of interpersonal communication and performance of virtual groups, including insights into the dependencies among group members and the influence of these dependencies on both communication and effectiveness. These results should have practical implications. Further, the project will modify and extend state-of-the-art machine learning tools in a way that should be useful for other social science investigations.