%0 Journal Article
%T 基于词典和规则的土耳其语形态消歧系统实现<br>Implementation of a Lexicon and Rule-Based Morphological System for Turkish Text
%A 张贵林
%A 易绵竹
%A 李宏欣
%A 李建
%A 易晓宇
%J Modern Linguistics
%P 1008-1017
%@ 2330-1716
%D 2021
%I Hans Publishing
%R 10.12677/ML.2021.94136
%X 本文提出了一种基于形态分析词典和上下文环境约束规则的土耳其语形态消歧方法，通过文本预处理、命名实体识别、固定搭配识别、未登录词处理、形态分析和形态消歧共6个模块，构建了一个实用的土耳其语形态消歧系统。实验中，系统对随机选取的15份新闻文本测试数据进行处理，结果显示，与未加入消歧规则的基线系统相比，文本中78.57%的形态歧义得到了解决，形态句法标注准确率达96.84%，提高了1.7个百分点。<br />
This paper proposed a hybrid approach that solves morphological disambiguation problem based on a Turkish Frequency List lexicon and contextual constraint rules. On this methodology, we have created a practical Turkish morphological disambiguation system consisting of text preprocessing, named entity recognition, fixed collocation recognition, unknown word recognition, morphological parsing and morphological disambiguation, a total of six modules. In the test, 15 online news texts were randomly selected, and by combining constraint rules the system gets 96.84% of all the morphosyntax features correctly parsed on the test data. Compared with the baseline system without disambiguation rules, 78.57% of the morphological ambiguities in the text were resolved and the accuracy increased by 1.7%.
%K 形态消歧，土耳其语，黏着语，混合方法<br>Morphological Disambiguation
%K Turkish
%K Agglutinative Language
%K Hybrid Approach
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=44352