太赫兹科学与电子信息学报, 2017, 15 (2): 279, 网络出版: 2017-06-06
基于语义角色分析的事件抽取技术
Event extraction technology by semantic role analysis
摘要
利用语义角色分析的方法对动态新闻进行事件抽取研究。通过对句子进行论元结构标注,抽取出句子中以谓语动词为中心的论元结构,将其转化为具体的语义角色,并与事件要素进行匹配,完成事件抽取工作。论文提出并重点研究了如何利用 VerbNet和 SemLink资源对动词的论元结构与事件要素进行匹配。抽取系统基于该方法对 1 000篇新闻语料进行了事件抽取实验,结果表明该方法的 F值达到 70.6%,具有一定的应用价值。
Abstract
The event extraction from news on the internet is performed by a method of semantic role analysis. The sentence in the news is annotated with arguments labeler; the argument structures centered with verbs are extracted and converted to a specific semantic role of the verb; and then the semantic roles are matched to the event elements. How to use VerbNet and SemLink resources to match the verb’s arguments and event elements is put forward and studied in detail. The experiment is carried out on 1 000 news corpus crawled from the web, and the results show that the F value is up to 70.6% and the proposed method has certain application value.
章顺瑞, 骆陈. 基于语义角色分析的事件抽取技术[J]. 太赫兹科学与电子信息学报, 2017, 15(2): 279. ZHANG Shunrui, LUO Chen. Event extraction technology by semantic role analysis[J]. Journal of terahertz science and electronic information technology, 2017, 15(2): 279.