A Model for Anticipatory Event Detection

Event detection is a very important area of research that discovers new events reported in a stream of text documents. Previous research in event detection has largely focused on finding the first story and tracking the events of a specific topic. A topic is simply a set of related events defined by user supplied keywords with no associated semantics and little domain knowledge. We therefore introduce the Anticipatory Event Detection (AED) problem: given some user preferred event transition in a topic, detect the occurence of the transition for the stream of news covering the topic. We confine the events to come from the same application domain, in particular, mergers and acquisitions. Our experiments showed that classical cosine similarity method fails for the AED task, whereas our conceptual model-based approach, through the use of domain knowledge and named entity type assignments, seems promising. We show experimentally that an AED voting classifier operating on a vector representation with name entities replaced by types performed AED successfully.