摘要
随着互联网的广泛应用,聚集了海量的文本数据,内含丰富的时空信息。面向文本的时空信息挖掘成为当前地理信息科学(Geographical Information Science, GIS)的研究热点,不断涌现出各种以自然语言处理(Natural Language Processing, NLP)技术为基础的挖掘方法。面向政务文本数据,对其进行事件级的时空模型建模,目的实现政务文本数据在事件级的简单分类、构建文本数据的结构化表达模型、探究事件级中时间空间的关系。实验证明,该模型对比之前的事件模型能更好地概括政务文本中的关键信息,针对于政务文本的结构化表达效果理想。
With the wide application of the Internet, a large amount of text data is gathered, which contains rich time and space information. Text-oriented spatio-temporal information mining has become a research hotspot in the current Geographic Information Science(GIS), and various mining methods based on Natural Language Processing(NLP) technology are emerging. This paper is oriented to government text data, and carries out event-level space-time model modeling. The purpose is to realize simple classification of government text data at the event level, construct a structured expression model of text data, and explore the relationship between time and space in event level. Experiments show that the model compares the previous event model to better summarize the key information in the government text, and the structural expression of the government text is ideal.
作者
李小龙
尹涵
LI Xiaolong;YIN Han(Faculty of Geomatics,East China University of Technology,330013,Nanchang,PRC)
出处
《江西科学》
2019年第6期958-963,988,共7页
Jiangxi Science
基金
国家重点研发计划项目(2017YFB0503700)
国家自然科学基金项目(41501437)
关键词
时空信息模型
政务文本挖掘
非结构化数据
事件信息抽取
spatio-temporal information model
government text mining
unstructured data
event information extraction
