ISSN 1000-1239 CN 11-1777/TP

计算机研究与发展 ›› 2021, Vol. 58 ›› Issue (11): 2475-2484.doi: 10.7544/issn1000-1239.2021.20200627

• 人工智能 • 上一篇    下一篇

融合上下文信息的篇章级事件时序关系抽取方法

王俊1,3,史存会1,3,张瑾2,俞晓明1,刘悦1,程学旗2,3   

  1. 1(中国科学院计算技术研究所数据智能系统研究中心 北京 100190);2(中国科学院网络数据科学与技术重点实验室(中国科学院计算技术研究所) 北京 100190);3(中国科学院大学 北京 100049) (wyswangjun@163.com)
  • 出版日期: 2021-11-01
  • 基金资助: 
    国家自然科学基金面上项目(91746301,61772498);国家重点研发计划项目(29198220,2017YFC0820404)

Document-Level Event Temporal Relation Extraction with Context Information

Wang Jun1,3, Shi Cunhui1,3, Zhang Jin2, Yu Xiaoming1, Liu Yue1, Cheng Xueqi2,3   

  1. 1(Data Intelligence System Research Center, Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190);2(CAS Key Laboratory of Network Data Science and Technology (Institute of Computing Technology, Chinese Academy of Sciences), Beijing 100190);3(University of Chinese Academy of Sciences, Beijing 100049)
  • Online: 2021-11-01
  • Supported by: 
    This work was supported by the General Program of the National Natural Science Foundation of China (91746301, 61772498) and the National Key Research and Development Program of China (29198220, 2017YFC0820404).

摘要: 事件时序关系抽取是一项重要的自然语言理解任务,可以广泛应用于诸如知识图谱构建、问答系统等任务.已有事件时序关系抽取方法往往将该任务视为句子级事件对的分类问题,而基于有限的局部句子信息导致其抽取的事件时序关系的精度较低,且无法保证整体时序关系的全局一致性.针对此问题,提出一种融合上下文信息的篇章级事件时序关系抽取方法,使用基于双向长短期记忆(bidirectional long short-term memory, Bi-LSTM)的神经网络模型学习文章中事件对的时序关系表示,再利用自注意力机制融入上下文中其他事件对信息,从而得到更丰富的事件对时序关系表示用于时序关系分类.通过TB-Dense(timebank dense)和MATRES(multi-axis temporal relations for start-points)数据集的实验表明:此方法能够取得比当前主流的句子级方法更佳的抽取效果.

关键词: 事件时序关系抽取, 时序关系分类, 事件关系识别, 自注意力, 双向长短期记忆

Abstract: Event temporal relation extraction is an important natural language understanding task, which can be widely used in downstream tasks such as construction of knowledge graph, question answering system and narrative generation. Existing event temporal relation extraction methods often treat the task as a sentence-level event pair classification problem, and solve it by some classification model. However, based on limited local sentence information, the accuracy of the extraction of temporal relations among events is low and the global consistency of the temporal relations cannot be guaranteed. For this problem, this paper proposes a document-level event temporal relation extraction with context information, which uses the neural network model based on Bi-LSTM (bidirectional long short-term memory) to learn the temporal relation expressions of event pairs, and then uses the self-attention mechanism to combine the information of other event pairs in the context, to obtain a better event temporal relation expression for temporal relation classification. At last, that event temporal relation expression with context information will improve the global event temporal relation extraction by enhancing temporal relation classification of all event pairs in the document. Experiments on TB-Dense (timebank dense) dataset and MATRES (multi-axis temporal relations for start-points) dataset show that this method can achieve better results than the latest sentence-level methods.

Key words: event temporal relation extraction, temporal relation classification, event relation identification, self-attention, bidirectional long short-term memory (Bi-LSTM)

中图分类号: