基于跨媒体分析的突发事件检测及趋势研判研究

突发事件的发生对社会经济产生重要影响,如何实时检测突发事件并对事件发展进行追踪,为应急处置和应急管理争取有效处理时间并提供决策支持具有重要研究意义.本文提出了一个基于跨媒体分析的突发事件检测模型,并对检测到的突发事件进行趋势分析和追踪.该方法通过对跨媒体文本进行深入分析,利用文本挖掘得到突发事件特征,并利用数据挖掘方法进行建模,有效提取突发事件;进一步地,利用事件分析和预测方法,对事件的进一步发展进行追踪,预测突发事件的未来走势.实证结果表明,本文提出的模型具有较好的检测精度,是突发事件检测与分析的新方法,具有较好的应用价值.

Emergency event detection and analysis for emergence management based on cross-media analytics

As emergency events have heavy impact on society and economics, how to detect and track the emergency events timely and then provide quick response and decision support to decision maker are critical important. In this paper, a cross-media analytics method is proposed for emergency event detection and analysis. A data mining method is offered to extract the opinion for single data source, and then a combined strategy is applied to integrate the results for event detection purpose. Furthermore, a model integration strategy is suggested to predict the trend of emergency event. Our proposed method is applied to an emergence scenario to validate the effectiveness. The experimental results show the effectiveness of the proposed method, comparing with baseline models. The finding reveals our proposed method is a potential one for emergency event detection and analysis.