基于最短路径的二元语法中文词语粗分模型的研究  

Research on A Chinese Words Rough Segmentation Model Based on Shortest-Path and Bi-gram Model

作　　者：甘秋云[1,2] 

机构地区：[1]福州海峡职业技术学院,福州350014 [2]福建工程学院国脉信息院,福州350014

出　　处：《现代计算机》2013年第17期7-10,共4页Modern Computer

摘　　要：词语粗分是分词后续处理的基础和前提,直接影响到分词系统最终的准确率和召回率。针对目前常用分词方法单一使用时存在的不足,综合机械分词的高效性和统计分词的灵活性,设计一种基于最短路径的二元语法中文词语粗分模型。实验结果表明,此粗分模型无论在封闭测试和开放测试中,还是在不同粗分模型对比测试和不同领域的开放测试中,都有较好的句子召回率。The words rough segmentation is the foundation and premise of the segmentation following pro- cessing, which directly influences to the final rate of accuracy and recall of the segmentation system.In view of the limitation of present segmentation methods when they are used singly, and to synthesize efficiency of mechanical segmentations and flexibility of the statistical segmentations, designs a Chinese words rough segmentation model which is based on shortest-path and Bi-gram model.Experimental results show that the model has good effect of coarse points and sentence recall rate, solve the most of ambiguity problems and decrease the problems of sparse data.

关 键 词：中文分词 最短路径 二元语法 词语粗分模型 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]