基于语料库的事件知识图谱构建与应用

Event oriented knowledge graph construction and application

人们以事件的形式认知客观世界,海量的非结构化信息不易于人们对事件的理解。传统的知识图谱关注实体及其之间的关系,事件是动态的,是比实体粒度更大的知识表示形式。研究以事件作为知识表示基本单元,首先基于中文突发事件语料库CEC 2.0,抽取事件和事件关系,然后以事件作为节点,事件关系作为边构建面向事件的知识图谱,再使用Echarts可视化构建的事件知识图谱,最后,以事件知识图谱作为知识,提出使用规则对事件缺省进行要素补全的方法。实例表明基于语料库的事件知识图谱构建方法是可行的。从事件的角度采用知识图谱形式结构化表示事件语义信息,为面向事件的理论和应用研究提供理论和技术支撑。

Human recognize the objective world in the form of events.It is difficult for human to understand events from the massive unstructured information.The traditional knowledge map focuses on entities and their relationships.Events are dynamic and represent knowledge with larger granularity than entities.We took events as the basic unit of knowledge representation,on the basis of the Chinese Event Corpus(CEC)2.0,extracted events and event relations,and then took events as nodes and event relations as sides to construct event knowledge graph,and realized visualization with Echarts.Finally,we proposed a rule-based method to complete the missing information of events with the event knowledge graph.The examples indicate the proposed method is feasible.In this work,event semantic information is represented structurally with the style of knowledge graph,which provides theoretical and technical support for the research of event-oriented theory and application.