A scalable TADGLN LSTM framework for modeling bibliometric research drift and analyzing research trends

Author-level bibliometric research drift is a gradual change in research focus over a specific time period due to emerging technologies, paradigm shifts in science, or shifting societal needs. Author-level bibliometric research drift must be identified for the comprehension of scientific progress trends, consistency in primary research objectives, and optimization of resource utilization. This study suggests a systematic approach to identify and quantify the Author-level bibliometric research drift through the aid of scholarly articles, authors’ contributions, and evolving nature of keyword distributions. The proposed framework measures author-level bibliometric research drift rather than semantic evolution of scientific concepts. Using natural language processing and deep learning techniques, we track changes in the research topic and project them onto a latent semantic structure. We demonstrate that the proposed TADGLN-LSTM, outperforms traditional approaches in identifying Author-level bibliometric research drift. Dynamic GNNs, with their ability to capture evolving interactions between authors, topics, and publications, excel in capturing temporal relationships and structural changes in the research domain. The implications of Author-level bibliometric research drift are profound, impacting funding, collaboration networks, and innovations particular to specific disciplines. Our approach enables institutions and policy makers to quantify the trends of research domains, thereby ensuring perpetual relevance and strategic foresight, ultimately facilitating a more responsive and forward-looking research ecosystem.