面向时政新闻文本的事件标注语料库构建  

Construction of Event Annotation Corpus for Current Political News Texts

作　　者：王瑞民[1] 叶娅娟 张坤丽[1] 昝红英[1] 韩英杰[1] WANG Ruimin;YE Yajuan;ZHANG Kunli;ZAN Hongying;HAN Yingjie(School of Information Engineering,Zhengzhou University,Zhengzhou 450001,China)

机构地区：[1]郑州大学信息工程学院,河南郑州450001

出　　处：《郑州大学学报（理学版）》2023年第2期25-32,共8页Journal of Zhengzhou University:Natural Science Edition

基　　金：国家重点研发计划项目(2017YFB1002101);国家社科基金重大项目(17ZDA138);国家自然科学基金项目(62006211);河南省科技攻关项目(192102210260)。

摘　　要：在现有的新闻领域标注语料库研究的基础上,结合时政新闻文本的特点,构建了面向时政新闻文本的事件标注语料库(event annotation corpus for current political news,EACPN)。EACPN从事件元素、人物角色及事件子类别等多个层面对21455篇时政新闻进行标注,涵盖了128523个事件元素和17919个子类别,整体标注一致性达到85.9%。所构建的EACPN为今后的时政新闻文本事件抽取研究和事件知识图谱构建提供了数据基础。Based on the existing studies of annotated corpora in the news field,the event annotation corpus for current political news(EACPN)was constructed according to the characteristics of current political news texts.EACPN annotated 21455 current political news from multiple levels including event elements,character roles,and event subcategories,covering 128523 event elements and 17919 subcategories,with an overall annotation consistency of 85.9%.The constructed EACPN provided a data basis for future study on event extraction of current political news texts and the construction of event knowledge graphs.

关 键 词：时政新闻文本 事件标注 语料库 事件抽取 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]