Mining diversity subgraph in multidisciplinary scientific collaboration networks: A meso perspective

This paper proposes a framework to analyze the interdisciplinary collaboration in a coauthorship network from a meso perspective using topic modeling: (1) a customized topic model is developed to capture and formalize the interdisciplinary feature; and (2) the two algorithms Diversity Subgraph Extraction (DSE) and Constraint-based Diversity Subgraph Extraction (CDSE) are designed and implemented to extract a meso view, i.e. a diversity subgraph of the interdisciplinary collaboration. The proposed framework is demonstrated using a coauthorship network in the field of computer science. A comparison between DSE and Breadth First Search (BSF)-based subgraph extraction favors DSE in capturing the diversity in interdisciplinary collaboration. Potential possibilities for studying various research topics based on the proposed framework of analysis are discussed.