Quantitative Analysis of Scholars' Topic Switching Behavior in Computer Science: A Two- Dimensional Metric Approach

Abstract

Computer science (CS) plays a crucial role in addressing challenges across various sectors, including education, finance, and healthcare. Investigating scholars' topic switching behavior quantitatively is essential for understanding the driving forces behind technological development in CS. However, the continuous growth of scientific publications poses challenges in accurately tracking scholars' topic switching. To address this, this study utilizes a large-scale dataset from the Microsoft Academic Knowledge Graph (MAKG) to construct bibliographic coupling networks for 16,794 scholars, enabling the detection of their research topics using the Louvain algorithm. Subsequently, this study introduces a novel two-dimensional metric to quantify scholars' topic switching behavior: topic switching degree and topic switching probability. This metric enables the classification of scholars into four distinct types: Type A, Type B, Type C, and Type D. Our analysis reveals that the majority of scholars belong to Type B, characterized by high topic switching probability but low degree, correlating with the highest quantity, quality, and citations of publications. Scholars categorized as Type A and Type D follow, while those falling under Type C, indicating low topic switching probability but high degree, tend to have the lowest publications and citations. Employing these new two-dimensional indicators to measure scholars' topic switching behavior and exploring its relationship with research performance contribute to advancements in technology within the field of computer science.