Leveraging Multimodal Information for Improved Cross-Document Event Coreference Resolution

Multimodal information, including images, can significantly improve cross-document event coreference resolution, especially for challenging mention pairs with low semantic and discourse-level similarity.