Event Relation Recognition And Application Based On Event Knowledge Graph

The traditional knowledge graph focuses on entities and relationships between reduce the damage caused by emergencies.Human beings recognize and understand the real world in the unit of events.At present,events are mostly used as the basic unit to process unstructured text in the field of natural language processing,and there is often a certain correlation between events.How to identify the relationship between events from unstructured text has become an important task in the field of natural language processing.Current event relationship recognition methods include pattern recognition based methods,traditional machine learning methods and currently very popular deep learning methods.The research results of these methods are excellent.However,there are often a variety of event relationships between events.At present,these methods are only oriented to a single event relationship.There is less research on identifying multiple event relationships at the same time,which is not enough to meet the needs of upper-level applications to obtain multiple event relationships.The event knowledge map is constructed with the event class as the node and the event class relationship as the edge.Compared with the traditional knowledge map,the event knowledge map contains richer semantic information,which can well show the occurrence and evolution law of dynamic events in the objective world.This paper proposes a method of using event knowledge graph to map event relationships to identify event relationships,which can simultaneously identify various event relationships and indirect connections that may exist between events and events,which has certain significance for event-oriented subsequent application research.This paper mainly includes the following two parts:First,event correlation is judged by integrating event co-occurrence,time factor and environmental factor.To identify the event relationship based on the event knowledge graph,it is first necessary to determine whether there is a correlation between two events.According to the joint co-occurrence probability of the action element and the object element of the event and the inference rule,it is judged whether the difference window range of the environmental element and the time element between the two events is within the specified window range to calculate the event correlation,and combine the event correlation threshold to judge whether the two events are related.Second,event relationship recognition based on event knowledge graph.On the basis of the event correlation judgment work,if relevant,the event relationship identification work can be continued.This paper first uses the event knowledge graph to detect event instances,and matches the elements of the event instance with the common feature set of the corresponding elements of the event class in the event knowledge graph to determine the event class to which the two events belong and find the corresponding event class in the graph.If both events can find the corresponding event class in the constructed event knowledge graph,use the given event relation inference rules and the existing knowledge of the event knowledge graph to reason and map the event relation,and realize the event relation identification.