Safe Bets and Risky Propositions: Leveraging Rich Data to Understand Potential in Science Teams

Northwestern University, Evanston IL

Investigators

Abstract

Innovation in science and technology is often the province of teams. As knowledge becomes ever more specialized, teams can tackle complex problems requiring insight from multiple domains. However, teams made up of diverse experts often struggle to combine knowledge effectively. In fact, research shows that while promising in theory, in practice, diverse teams often fail to realize their potential. Diverse teams are crucial to society. We are on the cusp of major breakthroughs in fields like artificial intelligence and brain science that require diverse expertise. These breakthroughs require large public expenditures in teams of scientists working in these fields. How can we identify the ''safe bets'' and ''risky propositions''? Which mixes distinguish mediocre teams from the teams whose work will fuel decades of important discoveries? These are the questions we seek to answer. We develop methods for characterizing scientific expertise and markers of productive collaboration among scientists based on machine-learning analysis of the full text of the papers they have written. Previous work relies on structural markers, and has yet to peer into the ideas themselves. Applying recent developments in text analytics machine learning, the full text of publications and patents is used to characterize expertise and diversity in ways that otherwise would not be possible. These advances have scientific and practical value. Practically speaking, we need a better way to predict which teams are most likely to be productive, and invest in them. Beyond this application, being able to predict diverse teams most likely to succeed will help to address societal challenges involving teams in other contexts as well, such as healthcare, disaster response, cybersecurity, and the military. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.