MEAFE: Multimodal Entity Alignment with Feature Enhancement and Weighted Aggregation

To generate a multimodal joint representation of entities, MEAFE first obtains embeddings of each modality using its encoders and the proposed feature enhancement method. It then performs weighted aggregation to fuse the multimodal information, resulting in a comprehensive representation.