Chinese Temporal Expression Recognition Combining Rules with a Statistical Model

Traditional rule-based methods for recognizing Chinese temporal expressions present a lower recall rate and they cannot recognize the event-type Chinese temporal expressions, thus, we propose a new Chinese temporal expression recognition method through combining rules with a statistical model. Firstly, we divide Chinese temporal expressions into seven categories and use basic time units as the smallest unit of recognition to simplify the complexity of rule-making. Then, we use regular rules to recognize Chinese temporal expressions and label the training data automatically. Meanwhile, we label the event-type temporal expressions that rule-based method cannot recognize. Lastly, we use the labeled training data to learn a Conditional Random Fields model for Chinese temporal expression recognition. Experimental results show that our proposed method significantly reduces the amount of annotation work and effectively improves the recognition performance. The F1 value reaches 88.73%, which is higher than the rule-based method by 6.13%.