摘要
关联规则挖掘是社会安全事件分析的重要方法之一,用于发现事件属性项及事件间的潜在关联。该文分析了社会安全事件的时空特性,利用时空关联规则挖掘方法分析事件属性项间的时空关联。为解决现有时空关联规则挖掘方法需要事先指定挖掘时间区间的问题,提出一种非指定时间约束的时空关联规则挖掘方法,根据事件时间属性值和时间划分粒度为事件空间和专题属性项增加时间标识,用时间标识代替时间属性值,得到全时间域内带有时间指向性的关联规则挖掘结果。以全球恐怖主义事件数据库为数据源,对该方法进行验证,结果表明其具有一定的可靠性与实用性,能够为社会安全事件的分析与预测、快速响应与防范提供决策依据。
Association rules mining is one of the most important methods in analyzing social security events for discovering potential relevance between events and event′s properties.The spatial and temporal characteristics of social security events have been analyzed,and spatio-temporal association rule mining has been used to analyze the spatio-temporal association relationships between event′s properties.In order to solve the problem of existing spatio-temporal association rules mining algorithms that specified time interval has to be pre-required,a spatio-temporal association rules mining method without specified time constraint has been put forward.Event′s temporal property values are replaced by temporal stamps which are made according to the event′s temporal property and time partitioning granularity.Temporal stamps are marked to the spatial and thematic attribute items,so that the event spatial and thematic attribute items with temporal stamps could reflect its intrinsic temporal characteristic.Through this method,mining algorithms could run in full time domain and time-directional association rules between event′s location,the performing organization,event type and target type supported by probability could be obtained.Global Terrorism Database has been used to validate the usefulness of this method,the result proves that the method is reliable and practical,and could provide a reliable basis for decision making on analysis and prediction,rapid response and prevention of social security events.
出处
《地理与地理信息科学》
CSCD
北大核心
2016年第3期14-18,共5页
Geography and Geo-Information Science
基金
国家自然科学基金项目(41501446)
地理信息工程国家重点实验室开放基金课题(SKLGIE2015-M-4-3
SKLGIE2015-M-3-1)
