Investigating latent interactions beyond direct connections is critical in complex networks. Traditional graph structures often fail to capture complex relationships, especially in high-order interactions. To address this issue, we enhance hypergraph learning capabilities by treating nodes as self-hyperedges and propose the Self-Hypergraph Graph Isomorphism Network (SHGIN) model by extending the Graph Isomorphism Network to the hypergraph. Extensive experiments on real-world cooperative networks have been conducted to demonstrate the effectiveness of SHGIN. The results indicate that our model displays superior classification accuracy compared to traditional graph neural networks in Semi-supervised learning (SSL). Furthermore, it surpasses existing hypergraph neural network models in certain datasets, highlighting its effectiveness in capturing high-orders relationships in complex networks.