藏语自动分词中的数词识别方法研究  

Identification of Numerals in Tibetan Word Segmentation

作　　者：完么扎西[1] 尼玛扎西[2] 

机构地区：[1]青海师范大学民族师范学院,青海西宁810008 [2]西藏大学图书馆和现代教育技术中心,西藏拉萨850000

出　　处：《西藏大学学报（社会科学版）》2015年第5期96-104,110,共10页Journal of Tibet University

摘　　要：藏语自动分词技术是藏语自然语言处理的基础。文章通过分析藏语真实文本中的数词分类、数词词形以及数词结构等,提出了一种基于规则的识别方法。文章中的藏语数词识别思想为:在自动分词过程中,通过判断待切分的词(w)i和已切分的词(wi-1)来重新组合。经对小学一至六年级的数学藏文版教材及1500个含各类数词的句子语料进行测试后,数词的识别准确率达97.7%。Tibetan automatic word segmentation is the foundation in the processing of Tibetan natural language.An identification method based on the rule was proposed by analyzing the numeral classification, numeral wordform and numeral structure in the Tibetan true texts. The baseline of this method is recombination by judging theword（wi） to be split and the segmented word（wi-1） in the segmentation process. The content of mathematicaltextbooks of primary school in Tibetan and 1500 sentences with various numerals were tested and the accuracyrate of numeral recognition reached up to 97.7%.

关 键 词：藏语自动分词 藏语数词 藏语数词结构 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]