Insights into knowledge evolution based on semantic representation and dynamic visual analytics

In the field of knowledge science, understanding the structure and dynamic evolution of knowledge is essential for advancing disciplinary development and anticipating research trends. However, current methodologies lack a unified semantic framework for the structured representation of knowledge, which impedes the quantitative analysis of its evolution and limits the ability to uncover complex relationships among knowledge entities. To bridge these gaps, we propose a structured knowledge representation method based on semantic embedding, enabling a deeper and more consistent understanding of semantic relationships within knowledge units. Building on this foundation, we introduce the concept of knowledge transfer flow to quantitatively analyze and visualize the dynamic evolution of knowledge hotspots over time, revealing the underlying mechanisms that drive knowledge transformation. Furthermore, we develop the KnowFlowViz system, which leverages interactive visual analytics to uncover intricate structural patterns and evolutionary dynamics within knowledge systems, thereby supporting decision-making and guiding future research directions. Our study reveals that established knowledge domains (such as long-standing disciplines) tend to maintain their dominant positions, while newly emerging knowledge entities often preferentially connect with these domains to form interdisciplinary linkages. This phenomenon of advantage accumulation and preferential attachment accelerates the growth and recognition of newcomers. The findings underscore the importance of fostering a more equitable and inclusive knowledge network, and they support the development of policies that nurture emerging disciplines and sustain a diverse, vibrant knowledge ecosystem.