<p>Over the past few years, researchers have extensively studied graph-based spectral multi-view clustering and developed state-of-the-art algorithms that yield exceptional performances. However, the limitations of standard graph-based structures, which capture only pairwise relationships between samples and cannot represent complex relationships among data samples, result in the loss of semantic information. On the other hand, the common practice of averaging distinct views to produce a consensus weakens the inherent information in each view. To tackle these issues, we suggested multi-view hypergraph spectral clustering via optimal transport. First, we create a sparse representation-based hypergraph structure to achieve spectral clustering on each view. This structure has the ability to identify more intricate relationships between data samples. Next, we utilize the optimal transport, particularly the Wasserstein distance, to align each view with the consensus view, thereby preserving local and global data information. Additionally, we reduce the dimensionality of the eigenspace by learning a low-dimensional subspace representation on the Stiefel manifold, where orthogonality is inherently preserved. This technique gives us an intrinsic geometric interpretation of the hypergraph spectrum and a deep understanding of the underlying structure. Lastly, we perform experiments on four real-world datasets to show the superiority of the proposed method.</p>