Multi-modal entity alignment in hyperbolic space

Many AI-related tasks involve the interactions of data in multiple modalities. It has been a new trend to merge multi-modal information into knowledge graph (KG), resulting graphs (MMKG). However, MMKGs usually suffer from low coverage and incompleteness. To mitigate this problem, viable approach is integrate complementary other MMKGs. end, although existing entity alignment approaches could be adopted, they operate Euclidean space, representations can lead large distortion KG’s hierarchical structure. Besides, visual yet not well exploited. In response these issues, work, we propose novel approach, Hyperbolic (HMEA), which extends representation hyperboloid manifold. We first adopt Graph Convolutional Networks (HGCNs) learn structural entities. Regarding information, generate image embeddings using densenet model, are also projected hyperbolic space HGCNs. Finally, combine structure use aggregated predict potential results. Extensive experiments ablation studies demonstrate effectiveness our proposed model its components.