Overlapping community detection on attributed graphs via neutrosophic C-means

Detecting overlapping communities in attributed networks remains a significant challenge due to the complexity of jointly modeling topological structure and node attributes, the unknown number of communities, and the need to capture nodes with multiple memberships. To address these issues, we propose a novel framework named density peaks clustering with neutrosophic C-means. First, we construct a consensus embedding by aligning structure-based and attribute-based representations using spectral decomposition and canonical correlation analysis. Then, an improved density peaks algorithm automatically estimates the number of communities and selects initial cluster centers based on a newly designed cluster strength metric. Finally, a neutrosophic C-means algorithm refines the community assignments, modeling uncertainty and overlap explicitly. Experimental results on synthetic and real-world networks demonstrate that the proposed method achieves superior performance in terms of detection accuracy, stability, and its ability to identify overlapping structures.