ME3A: A Multimodal Entity Entailment framework for multimodal Entity Alignment

Yu Zhao, Ying Zhang, Xuhui Sui, Xiangrui Cai

Abstract: Highlights•Multimodal entity alignment is crucial for knowledge graph coverage complementation.•Our multimodal entity entailment enables fine-grained cross-KG interaction.•Our mapping multimodal into unified textual space alleviates modality heterogeneity.•Experiments with various modality combination settings verify effectiveness.