The NEDC-GTOPSIS Node Influence Evaluation Algorithm Based on Multi-Layer Heterogeneous Classroom Networks

Abstract

To address the deficiency in the analysis of individual students within existing research on in-classroom social networks and the constraints of traditional centrality metrics in identifying influential nodes, this paper presents the NEDC-GTOPSIS evaluation method for evaluating node influence in multi-layer heterogeneous networks. Initially, students' friendship, interaction, and attribute information are leveraged to compute neighborhood overlap and attribute similarity between nodes, to construct the Composite Relationship Network. Subsequently, the Seat Similarity Network is constructed by applying the Nearest-Neighbor Effective Distance Criterion to compute seat similarity across various class sessions among nodes. Finally, the structure characteristics of two networks serve as influence decision indicators, and the GRA-TOPSIS algorithm, based on the combined weight method, evaluates nodes' influence. Experiments demonstrate that, compared to traditional single-layer relational networks and classical algorithms, this method can more effectively assess influential student nodes.