<p>Multiplex network analysis offers a powerful framework for exploring complex systems by capturing multiple modes of interaction or relationship types among entities. This study focuses on community detection within multiplex networks, aiming to uncover tightly connected groups of nodes that reflect the network’s multidimensional structure. Traditional network models primarily depend on pairwise connections, often neglecting higher-order structures that involve interactions among more than two nodes. Consequently, relying on a single structural perspective fails to fully capture the global characteristics of complex networks. To overcome this limitation, we propose a novel approach that integrates both low-order and high-order connections across all network layers into a unified proximity matrix. Specifically, we linearly combine the adjacency matrices, Jaccard similarity matrices, and motif-based adjacency matrices from each layer to construct a comprehensive similarity representation. This integration allows our method to capture both local and global structural features effectively. To further enhance community cohesion and detection accuracy, we introduce an edge pruning strategy that removes low-similarity connections, thereby reducing noise and redundancy. Finally, we apply Spectral Clustering (SPC) to the refined proximity matrix to identify communities. Experimental results on both synthetic and real-world datasets demonstrate that our method significantly outperforms existing approaches in terms of accuracy and robustness in community detection.</p>