基于“位置微博”的热点事件挖掘

Hot event mining based on"Location Weibo"

每日有数百万带有地理位置信息的微博(以下简称“位置微博”),为了更好地利用这些数据,帮助用户第一时间发现事发地点现场信息。微博公司启动了基于“位置微博”的热点事件的相关研究,通过获取微博博文坐标,运用Web墨卡托投影格子聚合转化、博文切词计算词频、区域事件关键词提取、墨卡托格子密度事件评估、地图聚合显示等算法,创建了基于“位置微博”的地点事件挖掘系统。并在中国台湾莲花地震事件挖掘中得到了充分验证,不仅提升了热点事件的发现效率和现场用户发声率,还大大提高了相关部门快速了解灾情、定点实施救援的能力。

Weibo creates millions of messages with geographic location information every day,so we started the project of a certain location events based on these"Location Weibo".With obtaining the coordinates of Weibo posts,we use Web Mercator grid aggregation transformation,blog cutting word frequency calculation,regional event keyword extraction,grid event evaluation,map aggregation display and other algorithms to built a searching system of location events based on"Location Weibo".It has been fully verified in the mining of hot events in the LianHua earthquake in Taiwan,China,which not only improved the discovery efficiency of hot events and the voice rate of on-site users,but also greatly improves the ability of relevant departments to quickly understand the disaster situation and implement rescue at designated locations.