基于位置概率模型的哈萨克语人名识别  

KAZAKH PERSONAL NAME RECOGNITION BASED ON POSITION PROBABILITY MODEL

作　　者：冯鲸华[1] 古丽拉.阿东别克 吴守用[1] 王花[1] 

机构地区：[1]新疆大学信息科学与工程学院,新疆乌鲁木齐830046

出　　处：《计算机应用与软件》2010年第12期21-23,91,共4页Computer Applications and Software

基　　金：国家自然科学基金项目(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)

摘　　要：针对哈萨克语文本中人名构成特点,提出了一种基于位置概率模型的哈萨克语人名可信度计算方法,构建了一个哈萨克语人名识别系统。系统分为训练和识别两个模块,识别过程是:首先从训练语料中提取特征进行训练,得到一个特征训练模型,然后利用训练好的特征模型及少量的附加规则,对测试文本中的人名进行识别,实验结果表明此方法可行。Considering the formation characteristic of personal names in Kazakh text,this paper proposes a method based on position probability model for computing the credibility of Kazakh personal names,and a recognition system for Kazakh personal name is constructed.It consists of a training module and a recognizing module.The recognition process is as follows：At first,features are extracted from the training corpus and to be trained,and a feature training model is derived.Then,this trained feature model and a few additional rules are used to recognize the Kazakh personal names in testing text.The results of experiments show that this method is feasible.

关 键 词：位置概率模型 哈萨克语人名识别 实体名识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]