MuSAM: Mutual-Scenario-Aware Multimodal-Enhanced Representation Learning for Semantic Similarity | IEEE Journals & Magazine | IEEE Xplore