From Macro to Micro: Visualizing Scientific Structures with the SDCL-Framework

Abstract Purpose This study aims to propose a scientific structure visualization framework that systematically characterizes the structure of scientific knowledge and analyzes the evolutionary trajectories of disciplinary topics, thereby providing data-driven support for scientific research decision-making. Design/methodology/approach To effectively identify the macro- and micro-structures of scientific knowledge, this paper proposes an innovative framework – SDCL-Framework, which integrates the SBERT pre-trained model, density contour visualization, and Log-likelihood ratio (LLR) label extraction algorithm. The synergy among these three components supports a scientific structure analysis paradigm that combines semantic depth, structural precision, and algorithmic robustness. Based on Essential Science Indicators (ESI) highly cited papers, the SDCL-Framework is used to construct visual maps of basic research scientific structures at both macro and micro levels. Findings The SDCL-Framework can reveal the internal micro-features and evolutionary processes within scientific structures, providing more detailed structural and relational information than traditional BERTopic-based results. Research limitations Since this study uses only ESI Highly Cited Papers as the dataset to demonstrate the feasibility and effectiveness of our visualization framework, the results serve as a methodological demonstration and cannot be used as a definitive guide for policy or resource allocation. Practical implications By integrating the SBERT model, density contour visualization, and LLR label extraction algorithm, the SDCL-Framework offers an innovative and comprehensive method for analyzing scientific structures. It assists in identifying research trends, optimizing the allocation of research resources, and supporting scientific decision-making. Originality/value The SDCL- Framework creatively combines multiple techniques, achieving both semantic depth and structural precision. It uncovers the evolution of scientific research from macro and micro perspectives, demonstrating unique value in the analysis of scientific structures.