Event Similarity Calculation Based On The Topic

Event similarity takes an important role in the construction of event ontology, andestablishment of event knowledge base and information inquiry based on events. Thetraditional calculation of event similarity is based on event six section definition, involvingthe action, object, environment, assertion, language expression, time factor. In theory, thesimilarity of event on the same topic should be higher than it is on the different topic, thushierarchical structure about topic and event is one of the factors that affect the similaritycalculation of the event, but the traditional method does not include it into the event similaritycalculation, there are shortages.This paper calculates event similarity based on the hierarchy of event and topic and theelements information of the event. It puts forward a new event similarity calculation based onthe topic. This method uses vector model to represent topic and event. It considers thesimilarity about event and topic, similarity between events and similarity between events time,at the same time assigns the weight coefficient to obtain the final event similarity with linearcombination method. In order to clearly express hierarchy about story event and topic, thispaper also puts forward the organization model of topic. In the experiment the practice getslots of similarity results for every events through assigning weight coefficient different value,and then based on the result it compares the performance of the new method and thetraditional one. The final data shows that the new method can more accurately reflect theevents similarity on the same topic.