Multimodal Alignment and Learning studies how to effectively align and fuse information from different modalities (text, image, audio, etc.) to achieve cross-modal understanding and reasoning. This direction has significant application value in areas such as scientific data mining and knowledge graph construction.