ME3A: A Multimodal Entity Entailment framework for multimodal Entity Alignment
Current methods for multimodal entity alignment (MEA) primarily rely on entity
representation learning, which undermines entity alignment performance because of cross …
representation learning, which undermines entity alignment performance because of cross …
Relational message passing with mutual information maximization for inductive link prediction
X Liang, G Si, J Li, P Tian, Z An, F Zhou - The Journal of Supercomputing, 2025 - Springer
Inductive link prediction (ILP) in knowledge graphs (KG) is gaining significant attention,
focusing on predicting missing triples involving unseen entities during training. While …
focusing on predicting missing triples involving unseen entities during training. While …
Multimodal Graph-based Variational Mixture of Experts Network for Zero-shot Multimodal Information Extraction
Multimodal information extraction on social media is a series of fundamental tasks to
construct the multimodal knowledge graph. The tasks are defined to extract the structural …
construct the multimodal knowledge graph. The tasks are defined to extract the structural …