The paper addresses the challenges of Riemannian graph representation learning, particularly in capturing motif regularity and addressing numerical instability. It proposes a novel model called Motif-aware Riemannian Graph Representation Learning (MotifRGC), which includes a new type of Riemannian GCN (D-GCN) and motif-aware Riemannian generative-contrastive learning. D-GCN constructs a diverse-curvature manifold using a product layer with a diversified factor, replacing the exponential/logarithmic map with a stable kernel layer. Motif-aware Riemannian generative-contrastive learning captures motif regularity through a min-max game in the constructed manifold. The model is evaluated on various datasets and shows superior performance compared to existing methods, demonstrating its effectiveness in learning node representations and capturing motif regularity.The paper addresses the challenges of Riemannian graph representation learning, particularly in capturing motif regularity and addressing numerical instability. It proposes a novel model called Motif-aware Riemannian Graph Representation Learning (MotifRGC), which includes a new type of Riemannian GCN (D-GCN) and motif-aware Riemannian generative-contrastive learning. D-GCN constructs a diverse-curvature manifold using a product layer with a diversified factor, replacing the exponential/logarithmic map with a stable kernel layer. Motif-aware Riemannian generative-contrastive learning captures motif regularity through a min-max game in the constructed manifold. The model is evaluated on various datasets and shows superior performance compared to existing methods, demonstrating its effectiveness in learning node representations and capturing motif regularity.