Cross-modal fine-grained alignment and fusion network for multimodal aspect-based sentiment analysis | Litlas