Cross Document Event Coreference

Cross-document event coreference resolution (CDECR) aims to identify mentions of the same real-world event across multiple documents, a crucial task for information extraction and knowledge base construction. Current research focuses on improving model accuracy by leveraging richer contextual information, including discourse structure, semantic relationships, and even figurative language, often employing large language models (LLMs) in conjunction with smaller, task-specific models or through innovative data augmentation techniques. Challenges include the inherent complexity of the task, particularly with diverse linguistic styles and the need for more robust and efficient annotation methods. Advances in CDECR will significantly improve the accuracy and efficiency of information retrieval and knowledge graph construction across various domains.