K-core sparsification for intracommunity edges (KSI) for efficient and scalable community detection in complex networks

Abstract

Community detection in complex networks often suffers from high computational cost and degraded community quality when graph reduction techniques excessively remove structurally important connections. To address this issue, this study proposes K-Core Sparsification for Intracommunity Edges (KSI), a community-aware graph reduction framework that integrates k-core-based backbone preservation with intra-community edge pruning to simplify network structures while maintaining community-relevant information. The proposed method was evaluated on benchmark networks with diverse structural characteristics using structural preservation, community quality, and computational efficiency metrics. Experimental results demonstrate that KSI preserves structural characteristics, improves modularity, reduces conductance, and achieves competitive computational efficiency compared with representative sparsification methods. These findings establish KSI as a community-aware structural refinement framework for scalable and effective community detection in complex networks.