Improving the Accuracy of Community Detection on Signed Networks via Community Refinement and Contrastive Learning

TL;DR

This work tackles the vulnerability of community detection on signed networks to noisy edge signs that create misaligned edges and distorted communities. It introduces ReCon, a model-agnostic post-processing framework that iteratively refines initial CD results through structural refinement, boundary refinement, contrastive learning, and clustering. Evaluations on 18 synthetic and 4 real networks with four baseline CD methods show consistent improvements in ARI and modularity, with ablations confirming the contribution of each component. The approach offers a practical, plug-in solution for more reliable CD in networks with noisy signs and is accompanied by publicly released code for reproducibility.

Abstract

Community detection (CD) on signed networks is crucial for understanding how positive and negative relations jointly shape network structure. However, existing CD methods often yield inconsistent communities due to noisy or conflicting edge signs. In this paper, we propose ReCon, a model-agnostic post-processing framework that progressively refines community structures through four iterative steps: (1) structural refinement, (2) boundary refinement, (3) contrastive learning, and (4) clustering. Extensive experiments on eighteen synthetic and four real-world networks using four CD methods demonstrate that ReCon consistently enhances community detection accuracy, serving as an effective and easily integrable solution for reliable CD across diverse network properties.

Figures (3)

• Figure 1: Overview of ReCon.
• Figure 2: CD accuracy and misaligned edge rates under increasing noise rates.
• Figure 3: Visualization of FEC and SPONGE on Rainfall.