基于动态时间窗口的突发监测研究

Research on Burst Detection Based on the Dynamic Time Window

针对Kleinberg突发监测算法中固定全局概率的缺陷,在动态时间窗口内计算突发词的基础概率,利用基于动态时间窗口的Kleinberg突发监测算法计算词的突发权重。结果表明利用环境概率计算得出的突发曲线更为精细,可识别出全局概率无法识别的较小的突发。

Aiming at the deficiency of fixed global probability of Kleinberg burst detection algorithm, the paper calculates the basis probability of burst topics in a dynamic time window, then calculates the burst weight using Kleinberg burst detection algorithm based on dynamic time window. The result shows that the burst curve calculated by environmental probability is more detailed, the smaller burst can be detected which global probability couldn＇t detect.