摘要
词性标注在自然语言信息处理领域中扮演着重要角色,是句法分析、信息抽取、机器翻译等自然语言处理的基础,对于哈萨克语同样如此。在基于词典静态标注的基础上分析了隐马尔科夫模型HMM(H idden M arkovModel)模型参数的选取、数据平滑以及未登录词的处理方法,利用基于统计的方法对哈萨克语熟语料进行训练,然后用V iterb i算法实现词性标注。实验结果表明利用HMM进行词性标注的准确率有所提高。
Part-of-speech（POS） tagging plays a key role in natural language information processing.It is the basis of natural language processing including syntactic parsing,information retrieval,and machine translation,etc.,and the same for Kazak as well.In the thesis we analyse the selection of HMM model parameters,data smoothing and the processing approach for new words based on static tagging on dictionary,and use statistics-based means to train mature Kazak corps;then we adopt the Viterbi algorithm to implement part-of-speech tagging.Experimental results show that the preciseness of POS tagging is improved with the use of HMM.
出处
《计算机应用与软件》
CSCD
北大核心
2012年第2期31-33,共3页
Computer Applications and Software
基金
国家自然科学基金(60763005)
国家教育部
国家语委民族语言文字规范标准建设及信息化科研项目(MZ115-92)
关键词
隐马尔科夫模型
哈萨克语
词性标注
自然语言处理
Hidden Markov model Kazak Part-of-speech tagging Natural language processing
