A New Method for Text Summary Extract Base on Event Network ⋆

Text can be partitioned and expressed by events and let events as the basic semantic unit to establish event ontology. Event network direct diagram constructed according to the relationship between events, which can better describe semantic information of the text and the importance degree of relationship between events. PAGERANK algorithm is employed to calculate and rank the event importance degree corresponding to each node in network diagram to output the original statement corresponding to events as abstract according to the time sequence of events. The experimental results show that extraction method of automatic summary based on the event network has achieved better performance.