Exploring Conceptual Structure: Network Analysis of Abstraction Levels Through Centrality Metrics



Author Information

Ekaterina Andronova, HSE University, Russia
Anastasiia Kapuza, HSE University, Russia

Abstract

Educational research has long explored the organisation of knowledge through conceptual structures, focusing on its features such as hierarchy and interconnections. The role of individual concept properties, such as their level of abstraction, remains underexplored despite its importance for developing a coherent understanding of a subject and constructing well-structured conceptual knowledge. However, assessing levels of abstraction empirically remains a challenge due to the lack of validated tools. This study applies network analysis to examine whether centrality metrics can differentiate levels of abstraction in conceptual maps. Drawing on Vygotsky’s view of concepts as dynamic elements within systems of meaning, we investigated whether a concept’s structural position reflects its abstraction level. A dataset of 242 concepts from concept maps of experts and students in data analysis was analysed. Concepts were categorised as superordinate, basic, or subordinate using formal definitions and expert judgement. Network metrics, including betweenness centrality, normalised betweenness centrality, and degree centrality, were calculated and tested using ANOVA and Tukey’s post-hoc comparisons. Superordinate concepts consistently showed higher betweenness and normalised betweenness centrality, indicating their bridging role within the conceptual structure. Degree centrality differentiated all abstraction levels when based on formal classifications but only distinguished subordinate concepts when based on participant maps. Superordinate concepts showed the highest overall connectivity. Network analysis, particularly the use of centrality metrics, provides a promising method for identifying abstraction levels in conceptual structures. These findings support new approaches to tracking conceptual development and designing tools that help learners build more coherent and abstract knowledge.


Paper Information

Conference: BCE2025
Stream: Assessment Theories & Methodologies

The full paper is not available for this title


Virtual Presentation


Comments & Feedback

Place a comment using your LinkedIn profile

Comments

Share on activity feed

Powered by WP LinkPress

Share this Research

Posted by James Alexander Gordon