A Span-based Multimodal Variational Autoencoder for Semi-supervised Multimodal Named Entity Recognition

30620

Knowledge Triples