Measuring Interdisciplinarity in Geology: A Semantic Analysis Approach

Interdisciplinarity is essential for addressing complex scientific problems that transcend disciplinary boundaries. Geology leverages methods from diverse domains to drive research innovation. However, quantitative evaluations of interdisciplinary connections between geology and other domains are lacking. Therefore, this study employed bibliometrics and natural language processing to assess the interdisciplinary trajectory of geology by analyzing temporal patterns in citation flows. First, a data set of geology‐related publications and their cited references was collected from the Scopus database. Then, a semantic text classification approach, integrating sentence transformers and cosine similarity, was implemented to categorize cited references into eight scientific domains: mathematical and physical science, chemical science, life science, engineering and materials science, earth science, information science, management science, and health science. Longitudinal analysis of the distribution of references across these domains reveals trends in interdisciplinary collaboration over time. Finally, N‐gram frequency analysis was performed on reference data associated with high‐growth domains to identify specific influential techniques bridging disciplines. The results demonstrate a pronounced increase in interdisciplinarity between geology and information science, especially in applications of artificial intelligence, since 2016, with an average interdisciplinary potential of 0.1484. Key techniques driving this integration include artificial neural networks, logistic regression, support vector machines, etc. Additionally, the eight domains were expanded into 126 sub‐disciplines to enable more detailed interdisciplinary analysis. Furthermore, three large language models were employed to verify the reliability of the adopted semantic analysis. The results suggest that our methodology provides robust approaches for quantifying interdisciplinary dynamics and is generalizable to other interdisciplinary fields.