基于辅助短语标记的名词短语识别

Recognition of Chinese noun phrase based on auxiliary phrase mark

名词短语的识别是自然语言处理领域中非常重要的子任务。而名词短语的识别性能与识别效率一直是研究人员关注的焦点,为了达到兼顾二者的目的,提出了一种基于辅助短语标记识别名词短语的方法。首先,在分析了短语不同分类体系的基础上,构建了一种映射公式,并根据该公式对不同分类体系的短语类别之间进行映射。然后,根据映射结果及短语的概率分布进行辅助短语标记的组合。实验结果表明,本文的方法在提高F值的基础上,有效地降低了系统的时间开销。

Noun Phrase Recognition is one of the most critical components in natural language processing field. The noun phrase recognition performance and its efficiency are the focus of researchers＇ attention. In order to combine the two elements, this paper proposes a method of recognizing noun phrases based on auxil- iary phrase mark. First, this paper presents a mapping between phrases by using the mapping formula based on the detailed analysis of the different classification system of the phrases. Then, according to the mapping results and the probability of the distribution of the auxiliary phrase mark, lots of combinations are estab- lished. Experimental results show that this method effectively reduces the time of noun phrase recognition without reducing the F-value.