Closer through commonality: Enhancing hypergraph contrastive learning with shared groups