Text Classification Method Based on Co-Occurrence Events

In view of the problem that emergency events are to be reported only following with the time sequence and theme types tend to be difficult to be distinguished from the texts. This paper first attempt to describe the reporting texts through events, in accordance with co-occurrence events appear in the texts, then endeavors to explore the properties such as distance and correlation between co-occurrence events in the text on the basis of the semantic relations between events to cluster the high similarity texts and eventually achieves text classification. Experiment shows that combining semantic relations between events with the relationship between co-occurrence events are more effective for text categorization.