Multi-Event Temporal Ordering by Event Order Ranking
Abstract: Extracting relationships and ranking the temporal order of document-level events is a challenging task in information extraction.
Previous methods primarily considered the event pair as the basic unit for processing, ignoring holistic connection among all events and background information remaining in the rest text.
To address these issues, we redefine the multi-event temporal ordering as Event Order Ranking(EORank) task, and introduce the Multi-Event Temporal Ranking(MEtR) model.
EORank simultaneously focuses on all events within a document from a holistic perspective.
We design order loss functions for MEtR, and our experimental results demonstrate their superior performance compared to other state-of-the-art models across EORank tasks of different settings.
Paper Type: Long
Research Area: Information Extraction
Research Area Keywords: event extraction, document-level extraction
Contribution Types: Model analysis & interpretability, Approaches low compute settings-efficiency, Surveys
Languages Studied: English
Section 2 Permission To Publish Peer Reviewers Content Agreement: Authors grant permission for ACL to publish peer reviewers' content
Submission Number: 39
Loading