基于混合模式的藏文音节切分

Tibetan Syllable Segmentation Based on Mixed Mode

通过藏文格助词的接续、结构以及上下文特征,提出基于规则、支持向量机、还原法等三层混合模式的藏文音节切分方法.藏文音节切分是藏文字频统计、分词、词性标注和机器翻译等研究领域的基础,其中藏文紧缩格歧义现象的正确识别、切分和还原是藏文音节切分的难点.经实验,混合模式藏文音节切分的F值为99.97%.

A Tibetan syllable segmentation method based on mixed mode of rules,support vector machine,restoration method was proposed through the analysis of case-auxiliary words and contextual features of Tibetan in this paper.The Tibetan syllable segmentation is the basis of many research fields such as Tibetan character frequency statistics,word segmentation,part-of-speech tagging and machine translation.Moreover,the correct identification,segmentation and restoration of Tibetan ambiguity case-auxiliary words are difficult points in Tibetan syllable segmentation.The experiment result showed that the F-measure score of 99.97%was obtained by using mixed mode Tibetan syllable segmentation.