《Table 1 Comparison between different event detection methods》

  提示：宽带有限、当前游客访问压缩模式

本系列图表出处文件名：随高清版一同展现

《A document-level model for tweet event detection》

1. 获取 高清版本忘记账户？点击这里登录

1. 下载图表忘记账户？点击这里登录

A comparison between two baseline methods and the proposed method is illustrated in Table 1.Both Twevent and Sensor consist of a burstiness-based feature selection before feature clustering and cluster filtering.However，Twevent and Sensor use different methods for burstiness calculation.Burstiness of Twevent is calculated as the difference between frequency in the current time window and the estimated average frequency in terms of standard deviation，while Sensor contrasts the frequency of the current time window with averaged history frequency without considering standard deviation.It is in line with Twevent by considering standard deviation.In addition，Twevent uses both history and future time window frequency for estimating burstiness，while Sensor only uses history information，which is more suitable for online event detection.The authors adopt online settings，using only history information.