Event Detection from News Articles

In this paper, we propose a new method for automatic news event detection. An event is a specific happening ill a particular time and place. We propose a new model in this paper to detect news events using a label based clustering approach. The model takes advantage of the fact that flews events are news clusters with high internal similarity whose articles are about an event in a specific time and place. Since news articles about a particular event may appear in several consecutive days, we developed this model to be able to distinguish such events and merge the corresponding news articles. Although event detection is propounded as a stand alone flews mining task, it has also applications in news articles ranking services.