The Michigan State University Construction Management program in the School of Planning, Design and Construction, or SPDC, was selected by the National Science Foundation, or NSF, to develop a research model of intelligent social network interventions.
This grant builds upon a grant received from NSF in 2018.
Project teams in the Architecture, Engineering and Construction, or AEC, industry are typically temporary and highly complex, multi-team systems.
They require smooth coordination and integration of ideas, while numerous individuals interact in a complex social network structure at sub-team and project team levels.
The objective of this $1.2M research project is to create a model to better enable individuals to develop the skills needed for working in these complex social systems, and provide short- and long-term economic and social benefits via improvements in student outcomes, individuals' skills and project outcomes.
This is a critical need for the AEC industry, as these project teams often have long-term social, economic and environmental impacts through their built environment products.
The main goal of the project is to offer a practical system, equipping individuals and organizations with sufficient means to facilitate multi-team coordination and project effectiveness.
“The results from this project will have a significant positive impact in the productivity of AEC workers that immediately take part in project teams, and will extend to a broad range of workforce via improvements in built environments,” said associate professor Sinem Mollaoglu, program director of the SPDC Construction Management program. “It will contribute to the science of organizations, engineering and research and development teams across industries that employ complex multi-team systems now and in the future.”
The trans-disciplinary team from MSU is made up of computer science, construction management, economics, education, engineering, organizational psychology and social networks experts. The team is led by Mollaoglu, and includes:
Kenneth Frank from the Department of Counseling, Educational Psychology and Special Education
Richard DeShon from the Department of Psychology
Jiliang Tang from the Department of Computer Science and Engineering
Hanzhe Zhang from the Department of Economics
Also, as part of the study, an outreach website will be developed to help train future workers. It will include new learning modules for project-based teaching and learning that incorporate intelligent social network interventions.
“This advanced use of technology will help people function more effectively as members of complex team structures that cross traditional organizational boundaries, and will support the development of critical teamwork skills in preparation for the future of work,” said Rick DeShon.
While social network analysis research has been carried out from various perspectives, little has been done to derive "actionable" insights and use these insights as intervention to improve communication, especially from the context of work.
This forms the basis for "dynamic network rewiring" based not only on human behavior, but also the work context, i.e., the goals of the work, via multiple cycles alternating between examining and intervening the network for behavior and context.
“This study epitomizes the land grant mission. It takes cutting-edge techniques in social network analysis, machine learning and team building, and applies them in real-time to the practical setting of construction management,” said Kenneth Frank. “Moreover, it has the potential not just to change current practice, but to help those in the field develop their capacity to better navigate complex social interactions in future settings."
To achieve these goals, the researcher team will use immediate and deep learning-enabled social network interventions to help individuals develop the skills needed for future of work and facilitate short- and long-term economic and social benefits.
The research team has formulated a longitudinal, comparative research design involving real-world AEC teams, as well as classroom — student-team test-beds — where equal numbers of cases are to receive manual, machine learning and no social network interventions.
Complementing the recent network intervention studies, this project focuses on complex and temporary multi-team systems.
The design will use multi-modal graph neural models to automate recognition of poor team functioning metrics so that problems can be diagnosed and interventions can be facilitated via augmentation of human cognition for multi-team coordination.
The design can accumulate knowledge obtained from past learning and adapt it for future learning, even in new domains.
“The success of this project has the potential to open doors and encourage new research opportunities in the direction of machine learning for effective teams," said Jiliang Tang.