摘要：
研究维吾尔语中的多音词现象，根据多音词的不同特点进行分类。利用词性和读音的映射关系消歧第1类多音词。根据词缀连接词干后是否发生元音弱化的特点消歧第2类多音词。提取上下文语境信息，使用最佳匹配读音的方法消歧第3类多音词。采用似然比方法进行关键词选择，并对不同窗口宽度的关键词选取方法进行对比实验。结果表明，该方法可以得到错误率为20.9%的多音词消歧效果。
关键词:
维吾尔语,
多音词消歧,
分类,
元音弱化,
最佳匹配读音,
关键词选取
Abstract:
This paper deeply investigates the homograph in Uyghur language and classifies them according to the different features of homograph, disambiguates the first type of homograph according to the mapping relation between the part of speech and pronunciation, disambiguates the second type of homograph according to vowel weakening when suffix attaches to a stem, and optimal pronunciation mapping method is used to disambiguate the third type of homograph by extracting the contextual features of homograph. Log-likelihood ratio is used to select keywords and keyword selection experiment of different window size is also conducted. Experimental result shows that the homograph disambiguation performance of can be got to 20.9% error rate through the research idea of this paper.
Key words:
Uyghur language,
homograph disambiguation,
classification,
vowel weakening,
optimal mapping pronunciation,
keyword selection
中图分类号:
姑丽加玛丽.麦提艾力a, 艾斯卡尔.孜b, 古丽娜尔.力a, 艾斯卡尔.木都拉a. 基于分类及最佳匹配读音的维吾尔多音词消歧[J]. 计算机工程, 2013, 38(18): 22-25.
GU Li-Jia-Ma-Li-.Mai-Chi-Ai-Li-a, AI Shi-Ka-Er-.Zi-b, GU Li-Na-Er-.Li-a, AI Shi-Ka-Er-.Mu-Dou-La-a. Uyghur Homograph Disambiguation Based on Classification and Optimal Mapping Pronunciation[J]. Computer Engineering, 2013, 38(18): 22-25.