论文阅读记录

MAVEN-ARG

  • 标题：MAVEN-ARG: Completing the Puzzle of All-in-One Event Understanding Dataset with Event Argument Annotation
  • 作者：Xiaozhi Wang(THU)
  • 发表会议/期刊：ACL
  • 年份：2024
  • 链接：MAVEN

Intro

  • Background:

    • event understanding is typically organized as three information extraction tasks: event detection (ED), event argument extraction (EAE), event relation extraction (ERE).
    • MAVEN
    • MAVEN-ERE
    • MAVEN-ARG

  • Motivation：A large-scale dataset covering all the event understanding tasks has long been absent.

Main

  • Event Understanding：
    Generalization

  • Method：

    • Event Schema Creation: 162 event types, 612 unique argument roles, and 14,655 words of definitions
      Generalization
    • Entity Annotation:annotate entities for the 4,480 MAVEN documents and annotate entity coreference.
    • Event Argument Annotation

Data Analysis

  • table 2:
    Generalization

  • figure 2
    Generalization

  • figure 3
    Generalization

  • main results:
    Generalization

  • others:

    • Experiment Results of Fine-tuned Models
    • Experiment Results of LLMs
    • Analysis on Trigger-Argument Distance
    • Analysis on Entity and Non-Entity Arguments
    • Error Analysis

  • Future Event Prediction Demonstration:
    Generalization