SMGT: A self-supervised multimodal graph transformer with coverage-aware selection for video summarization with nanotechnology-inspired nanoscale representation learning. ETN. 2026;10(3):1469-1484. doi:10.56053/10.3.1469