越南语短语树到依存树的转换研究

发布时间：2018-05-16 03:31

本文选题：句法分析 + 中心子节点过滤表　；参考：《计算机科学与探索》2017年04期

【摘要】：依存句法分析是自然语言处理的一个关键环节,目前对于越南语短语结构树的研究比较多,而依存结构树的研究就显得十分薄弱。提出了一种新的方法,尝试结合越南语的语言特点和语法特征,利用中心子节点过滤表的思想与统计的方法将越南语的短语结构树转换成依存结构树。首先依据中文依存关系标注体系与越南语的语法规则,制定出依存关系列表;然后结合越南语的语言特点,制定出中心子节点过滤表,利用中心子节点过滤表的思想进行初步转化;最后使用依存关系标注器来进行依存关系标注。基于转换后得到的依存结构树,利用MSTParser工具进一步训练得到更多的越南语依存结构树。对实验结果进行了抽样评估,树库转换的准确率达到了89.4%,较好地解决了越南语由短语树到依存树的转换问题。
[Abstract]:Dependency parsing is a key link in natural language processing. At present, there are many researches on Vietnamese phrase structure tree, but the research on dependency structure tree is very weak. In this paper, a new method is proposed to transform the phrase structure tree of Vietnamese into dependency structure tree by combining the linguistic and grammatical features of Vietnamese with the idea and statistics of the central sub-node filter table. Firstly, according to the Chinese dependency tagging system and the grammar rules of Vietnamese language, the dependency list is worked out, and then the central sub-node filter table is made according to the language characteristics of Vietnamese. The idea of filtering table of central child node is used to change the idea of the filter table, and the dependency tagger is used to label the dependency relationship. Based on the transformed dependency tree, the MSTParser tool is used to further train and obtain more Vietnamese dependency tree. The experimental results are evaluated by sampling, and the accuracy of tree-base conversion is 89.4. the problem of Vietnamese translation from phrase tree to dependency tree is well solved.
【作者单位】：昆明理工大学信息工程与自动化学院;昆明理工大学智能信息处理重点实验室;
【基金】：国家自然科学基金Nos.61262041,61363044,61472168 云南省自然科学基金重点项目No.2013FA030~~
【分类号】：TP391.1

【相似文献】