Article Preview
TopThe trend of utilizing data analysis techniques to record, process, and analyze classroom teaching data for individual students has emerged along with the integration of teaching and emerging information technology (Abu Talib et al., 2021; Rapanta et al., 2021). Van den Bos et al. (2018) used SNA, a data analysis method for studying students’ classroom dynamics, academic performance, and interaction patterns, to explore how individuals in central positions within network structures exhibit more prosocial behaviors and relational aggression. Zhao et al. (2021) employed diverse indicators from students’ collaboration networks to assess node network positions, determining that students in central positions within the network tend to experience a higher sense of belonging. Williams et al. (2019) utilized centrality measures within a classroom interaction network to longitudinally study student engagement, confirming the crucial role of student engagement in academic achievement.
In addition to mining the relationship between network structural features and student interactions, as well as academic performance, studies have shown that students’ seating arrangements contain valuable social information. Analyzing the distribution of students’ seating in the classroom, along with SNA, can help capture their social relationships. This approach sets the foundation for parsing complex student social behaviors (Minami & Ohura, 2020; Pei et al., 2018).
Existing classroom social network studies concentrated on single-layer relationship networks, failing to consider that a complete network system is formed by multiple layers of subnetworks with different relationship structures and heterogeneous attributes (Huang et al., 2021). Mapping information diffusion solely to a single-layer network may miss complex interaction patterns between nodes, leading to an inaccurate analysis of dynamic characteristics and topological features (Interdonato et al., 2020).
As research advances, effectively identifying influential nodes in MLH networks has become a challenging problem. Wan et al. (2022) explored constructing MLH networks from diverse relationships and structural features, and employed multi-relationship coupling information and transmission mechanisms to measure node importance. Li et al. (2023) introduced a general coupling feedback mechanism for node importance assessment in multi-layer networks, emphasizing the ability of inter-layer feedback to capture more details in real systems. Wang et al. (2017), from various types of relationships and network structural features, constructed an MLH network and used multiple relationship coupling information and transmission mechanisms to measure the importance of nodes.
Researchers in the field of SNA have introduced numerous algorithms to mine influential nodes within the network (Das et al., 2018, 2020; Singh, 2022). However, most centrality-based methods for evaluating the importance of nodes have limitations. Their adaptability and accuracy are easily influenced by factors like network structure and scale (Liu et al., 2021). To address this challenge, multi-attribute decision-making (MADM) methods can be extended to identify crucial nodes across various dimensions and comprehensively evaluate node influence using multiple centrality features (Dong et al., 2022).
Liu et al. (2023) utilized the TOPSIS algorithm to model node features and employed PageRank to measure the interdependence between nodes, assessing the contribution level of each node. Yang et al. (2018) proposed a dynamic TOPSIS method based on the GRA algorithm and SIR model to identify key nodes. In the research on classroom social networks, Shou et al. (2023) utilized students’ seat similarity to construct a classroom social network. They then employed the GRA-TOPSIS algorithm to unearth key student nodes with negative impact.