A novel method for online bursty event detection on Twitter

As one of the most popular social media platforms, Twitter has become a tool that people widely used to share their contents, their interests and events with friends. Meanwhile, we are facing a big challenge to find the bursty events from the large volume of continuous text streams quickly and accurately due to millions of data produced every day. In this paper, we proposed a BBW (Basic-Burst Weight) method based on the Time Window to extract bursty words, then we exploit these bursty words to detect the meaningful bursty events combined with hierarchical clustering algorithm. Our experiments on a large twitter dataset show that our method can detect bursty events timely and precisely.