Event detection in Twitter with an event knowledge base

We present two methods for event detection in Twitter using an event knowledge base. The knowledge base used contains world events reported in the media that we identify as multi-lingual clusters of mainstream news stories. Given this fact, we reduce the problem of event detection to matching tweets to mainstream news stories. The first method consists of using URLs to mainstream news sites present in tweets and in the knowledge base. We use this method to build a supervised corpus of tweets and then create and evaluate a supervised classifier as our second method. Experimental evaluation on real-world data shows that the proposed methods perfom well on our dataset.