XMolCap: Advancing Molecular Captioning Through Multimodal Fusion and Explainable Graph Neural Networks | IEEE Journals & Magazine | IEEE Xplore