【作者】 江荻；

【Author】 Jiang Di Department of Computational Linguistics

【机构】 中国社会科学院民族学与人类学研究所计算语言学研究室；

【摘要】 本文定义和描述了现代藏语句法组块的基本类型以及相关的形式标记,并在此基础上提出藏语自动分词的组块方法。而实施组块分词方法的措施包括按照一定顺序原则识别组块的形式标记,通过各类标记函数集、辅助词表,以及从组块中抽取的句法信息确定组块的边界,然后对组块进行分词和词性标注。进一步的设想是对组块进行归并,使其与藏语句法成分形成一致关系,达到消除嵌套组块和利于句法理解的目的。更多还原

【Abstract】 The present paper describes the basic types of syntactic chunks and their formal markers in modern Tibetan. And a scheme of automatic word-segmentation based on chunks has been provided according to the features of Tibetan syntactic structures. In order to carry out the scheme some primary work must be done as pre-processing, setting up small tables of formal markers of each chunk, the verbal paradigm, special tables of differentiating cases and homographs, etc.. Then with all these information we can find the left and right bounders of each chunk, and go on segmenting words with a dictionary and tagging within chunks.更多还原