Chinese Part-of-speech Tagging Based on Single Word Feature Templates
Yu Zhengtao · Journal of Shanxi University · 2011
Language model for practical application should take up less storage space and loading speed,so we studied the relations of the feature template set and the training model size,tagging accuracy for Chinese part-of-speech tagging by using maximum entropy model.The closed evaluations are performed on PKU,NCC and CTB corpus from the Bakeoff-2007,and the comparative experiments are performed on different feature templates.The experimental results show:the size of the model has greatly increased and tagging accuracy is not improved using combination of two-word feature templates,while the size of the model is 1/5 of the original size,but without any marked decline in accuracy using the single word feature template.