Measuring Research Convergence in Interdisciplinary Teams Using Large Language Models and Graph Analytics

A team of 11 researchers led by Wenwen Li from Arizona State University has published a novel framework that uses AI to solve a persistent challenge in science: understanding how interdisciplinary teams converge on shared knowledge. The system, detailed in the arXiv paper 'Measuring Research Convergence in Interdisciplinary Teams Using Large Language Models and Graph Analytics,' creates a multi-layer analytical approach. It first uses LLMs to extract and structure research viewpoints from team members, aligning them with the Needs-Approach-Benefits-Competition (NABC) framework. These structured viewpoints form a common semantic foundation for three complementary analyses: identifying popular and unique viewpoints, quantifying cross-domain influence using network centrality measures, and tracking the temporal flow of ideas to capture convergence dynamics.

To address the inherent uncertainty in LLM-based inference, the framework incorporates a crucial human-in-the-loop component. Expert validation is performed through structured surveys and cross-layer consistency checks, ensuring the AI's interpretations align with human judgment. The team demonstrated the framework's value through a case study on water insecurity in underserved communities as part of the Arizona Water Innovation Initiatives. The analysis revealed measurable increases in viewpoint convergence over time and identified specific patterns of domain-specific influence, proving the method's practical utility for managing and accelerating complex, collaborative research projects.