基于字向量的条件随机场的中文分词方法  

Chinese Word Segmentation Method based on Combination of Word Embedding and CRFs

作　　者：周寅 ZHOU Ying(Wuhan Institute of Shipbuilding Technology,Wuhan 430050,China)

机构地区：[1]武汉船舶职业技术学院

出　　处：《武汉船舶职业技术学院学报》2019年第4期148-152,共5页Journal of Wuhan Institute of Shipbuilding Technology

摘　　要：本文利用深度学习中的词向量来稍加改进,利用Word Embedding与条件随机场结合,然后进行线性组合,使分词达到更加高效。运用北大的语音资料库提供的实验语料进行实验,进一步证实实验结果,融合条件随机场的技术,在中文分词方面取得更优秀的效果,在北京大学提供的语料库的封闭测试中准确率和召回率都得到了较高的成绩。并且在开放式领域中也取得了明显的提高。In this paper,the word vector in deep learning is used,linear combination of Word Embedding and conditional random is carried out to make the segmentation more efficient.The experimental data provided by the speech database of Peking University is used in the experiment.The experimental results show that the combination of Word Embedding and conditional random field technology has achieved better results in Chinese word segmentation,and the accuracy rate and recall rate in the closed test of the corpus have achieved higher results.Significant improvements have also been made in the open field.

关 键 词：中文分词 CRFS 字向量 

分 类 号：H13[语言文字—汉语]