Mapping Academic Disciplines via Higher-Education Curriculum Data
Where
2024 KPS Spring Meeting
When
Tuesday-Friday, April 25
Where
P2-st.008
What
Basic Natural Science Undergraduate Curriculum Data
Who
GIM Gahyoun 1, SEO Jibeom 2, KIM Beom Jun 2, LEE Sang Hoon 1,*
(1Department of Physics, Gyeongsang National University, 2Department of Physics, Sungkyunkwan University)
How(Abstract)
To overcome possible limitations of journal-based analyses in scientometrics or science of science (SoS), this study proposes a framework that utilizes higher-education course data as an ideal resource for the established core part of human knowledge, while its growing cutting-edge part would correspond to the conventional journal-based SoS. We aim to quantitatively explore the connectivity and the structure of interdisciplinary foundational knowledge and to reveal the interconnections between knowledge concepts. First, we integrate the official Korean basic natural science curriculum data from the Ministry of Education in Korea. Although the raw data covers extensive sets of curricula in Korean universities, it is prohibitively hard to manually define a proper unit of courses due to multiple nomenclatures and different numbers of spanning semesters allocated to each course. For a systematic assignment of each unit of courses, we utilize other metadata in the curriculum data such as course descriptions, textbooks, etc. By employing a sophisticated pre-trained transformer model and the neural network too assess the interrelations and similarities, we conduct a community detection process, which results in the assemblage of similar subjects from a mesoscale perspective and rescaling each cluster as a (super)node. We analyze the correlation among the nodes by properly connecting them in the embedding space, e.g., in terms of the bipartite network composed of the coursework nodes and institution nodes. By inspecting the bipartite network and its projected co-occurrence networks, we quantitatively reveal the interconnectedness and structural characteristics of each discipline. The insights derived from our data and analytical method illuminate the intricate interplay among academic fields through fresh lenses, offering a quantified appreciation for the role of foundational knowledge that has been overlooked in SoS research focused on the new discovery represented by journal papers and patents.
Keyword:
Scientometrics, Curriculum Data, Co-occurrence Network