Dual Label-Guided Graph Refinement for Multi-View Graph Clustering

With the increase of multi-view graph data, clustering (MVGC) that can discover hidden clusters without label supervision has attracted growing attention from researchers. Existing MVGC methods are often sensitive to given graphs, especially influenced by low quality i.e., they tend be limited homophily assumption. However, widespread real-world data hardly satisfy This gap limits performance existing on homophilous graphs. To mitigate this limitation, our motivation is extract high-level view-common information which used refine each view's graph, and reduce influence non-homophilous edges. end, we propose dual label-guided refinement for (DuaLGR), alleviate vulnerability in facing Specifically, DuaLGR consists two modules named module encoder module. The first designed soft node features then learn a matrix. In cooperation with pseudo second module, these graphs refined aggregated adaptively different orders. Subsequently, consensus generated guidance label. Finally, encodes along produce iteratively clustering. experimental results show superior coping data. source code available at https://github.com/YwL-zhufeng/DuaLGR.