Intro

TLDR: This paper treats document-level argument extraction as a conditional generation problem by filling in event templates.

Argument Extraction Model Overview

Dataset

The WikiEvents datasets is under data/wikievents.

The RAMS dataset is for download here.

The ACE dataset is provided by LDC and unfortunately we cannot release it directly. See this link for details.

v0.1 (April 12, 2021): Basic generation model for argument extraction. (This does not include the post-processing script.)

This page is currently under construction.