Temporal Relation Recognition Between News Events Based On Statistics And Rules

The temporal relation between the atomic events described in the text is some kindof timing relation, which is a natural link between the events. The temporal relation in the linguistics includes synchronic and diachronic. The events with synchronic relation generally occur synchronicly, and the events of diachronic relation take palace in the order of time with a clear relationship.The event provides an effective semantic structure, in line with the human cognitive model. Therefore, the temporal relation recognition between events is actually a basicsemantic level research, and plays a key role in the field of natural language processingapplications.In this paper, the temporal relation of news atomic events is recognized based on statistics and rules, and the essence of the method is the combination of pattern matchingand machine learning. In the event temporal relation recognition is explicitly associatedby the timing function word. Hence, we recognize the temporal relation between the news events through the establishment of the timing function word semantic rule base. In recognition of the atomic event temporal relation implicitly by predicates,we mainly depends on the results of the statistics learning by the predicate co-occurrence correlation toget effectively identify recognition, and it is more convenient and efficient in the identification process. The experimental results show that the news event temporal relation recognition based on statistics and rules is effective and effeiciency, and the timing function word and predicate co-occurrence knowledge applies to the recognition of news events.