A1193
Title: Hierarchical clustering of subject co-occurrence networks: A robustness-based journal interdisciplinarity assessment
Authors: Wei-Chu Chiang - National Chengchi University (Taiwan) [presenting]
Frederick Kin Hing Phoa - Academia Sinica (Taiwan)
Hiroka Hamada - The Institute of Statistical Mathematics (Japan)
Abstract: Interdisciplinarity is central to contemporary science, yet its measurement depends critically on how disciplinary boundaries are defined. A network-based framework for mapping the global structure of human knowledge and assessing journal interdisciplinarity is proposed. Using journal-level subject classifications from Web of Science (WoS), a weighted subject co-occurrence network (SCN) is constructed in which nodes represent subjects and edges capture their co-occurrence within journals. Hierarchical clustering based on weighted shortest-path distances yields 20 higher-level subject groups, providing a transparent and updateable disciplinary map that balances aggregation and granularity across existing classification systems. Building on the SCN, a robustness-based interdisciplinarity index grounded in communication efficiency is introduced. By simulating the addition of journals with specific subject types, their marginal contribution to global network connectivity is quantified. The index exhibits a heavy-tailed distribution, indicating that a small number of subject combinations disproportionately enhance cross-disciplinary integration. Results further distinguish weak within-cluster interdisciplinarity from structurally significant cross-cluster integration. The framework offers a lightweight and structurally grounded alternative to citation-intensive approaches for large-scale interdisciplinarity assessment.
