Identification of core sub-team on scientific collaboration networks with Shapley method

Abstract

Identifying the core sub-teams that drive productivity in scientific collaboration networks is essential for research evaluation and team management. However, existing methods typically rank individual researchers by bibliometric impact or select structurally cohesive clusters, but rarely account for both collaboration patterns and joint scientific output. To address this limitation, we propose a novel two-dimensional framework that integrates network topology with research performance to identify core sub-teams. Specifically, we measure each sub-team’s marginal structural contribution using the Shapley value and quantify its collective impact using a sub-team H-index. To efficiently identify high-contributing sub-teams, we employ the Monte Carlo Tree Search algorithm, along with an approximation strategy to estimate Shapley values under computational constraints. We evaluate our method on 61 real-world scientific collaboration teams from Web of Science and Baidu Scholar data. Experimental results validate the effectiveness of our method in identifying core sub-teams, with the highest collaborative and citation impact. The proposed method offers a valuable analytical tool for research managers and funding agencies seeking to locate high-impact collaborative clusters, and it provides a generalizable framework for studies requiring the integration of structural and performance-based indicators in network analysis.