Analysis of the Impact of Researchers' Knowledge Diversity on the Research Productivity from a Knowledge Meta-Perspective

In the evolving landscape of interdisciplinary research, the researchers’ knowledge diversity plays a pivotal role in fostering innovation. This study proposes a novel method, merging deep learning with a knowledge meta-perspective, to gauge and depict researchers’ knowledge diversity. By leveraging the BioBERT algorithm, we extract and vectorize knowledge meta from researchers, measuring diversity through cosine similarity. Analyzing data from 16,245 pharmaceutical scientists and 1,846 library and information scientists, we explore how knowledge diversity impacts research productivity via Poisson regression. Findings reveal an inverted U-shaped relationship between researchers’ knowledge diversity and research productivity, moderated positively by their prior research citations. This suggests that researchers should find a balance between expanding their broad disciplinary backgrounds and deepening their areas of specialization in the research process, so that their knowledge diversity is truly reflected in high-quality research productivity with a high level of cross-cutting knowledge. Strengthening cross-institutional, national cooperation, while support and incentives for seasoned researchers can sustain their contributions to research endeavors.