基于微博文本的灾害事件信息时空过程聚合方法

Spatio-temporal process based information aggregation method of disaster events in microblog text

各类灾害事件频发已成为全球可持续发展面临的重大威胁。在大数据环境下,微博文本逐渐被应用于灾害管理的预防、准备、响应和恢复工作。以往研究多关注微博文本中灾情信息的获取,却忽略对这些碎片化信息进行有序化整合。本文从时空视角构建多层次的灾害事件信息模型,在抽取出微博文本中灾害事件信息要素的基础上,提出基于“对象-状态”的过程信息聚合方法,解决微博文本中灾害事件信息分散化、时空粒度多样化和无序化的问题。基于新浪微博进行台风“利奇马”事件的案例分析,结果表明,本文方法能够全面地获取灾害事件过程中各个时空节点上的灾情信息,有利于从微博文本中挖掘小尺度下的灾害突发状况。

The frequent occurrence of various disaster events has become a major threat to global sustainable development.In the era of big data,microblog text has been gradually applied to the prevention,preparation,response and recovery of disaster management.However,previous studies mostly focused on the acquisition of different disaster information in microblog text,while ignoring the structural integration of this fragmented information.In this paper,an information aggregation method of disaster events in microblog text is proposed.First,a multi-level disaster event information model is constructed from the spatio-temporal perspective.Then,the information elements of disaster events are extracted from the microblog text.Finally,a process-oriented information aggregation model based on“Object-State”is constructed.On this basis,it can be solved that the problems of decentralization,spatio-temporal granularity diversification and disorder of disaster event information in microblog text.Based on the case study of typhoon“Lekima”event using Sina Weibo,the results show that the method proposed in this paper can comprehensively acquire the information of different objects on every spatio-temporal node during the process of typhoon events,which is beneficial to mining disaster emergencies in small scale from microblog text.