Query-centric cross-document event timeline extraction is the task of assembling a timeline of events from multiple documents, shaped by a particular query focus. The arXiv paper 'TimelineAgent' proposes an LLM-based framework for this task. The abstract describes the problem as challenging due to heterogeneous event modeling and reasoning.
Because the source is only the paper's abstract, details about how TimelineAgent works or how it was evaluated are not available. The abstract emphasizes the difficulty of handling diverse query focuses and the need for reasoning across documents. The paper is listed as arXiv:2610.25733v1.