基于条件随机场的维吾尔文机构名识别  

Uyghur organization name recognition based on conditional random fields

作　　者：买合木提.买买提 王路路 吐尔根.依布拉音[1,2] 艾山.吾买尔 卡哈尔江.阿比的热西提[1,2] Maihemuti Maimaiti;WANG Lu-lu;Tuergen Yibulayin;Aishan Wumaier;Kahaerjiang Abiderexiti(College of Information Science and Engineering,Xinjiang University,Urumqi 830046,China;Xinjiang Laboratory of Multi-Language Information Technology,Xinjiang University,Urumqi 830046,China)

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 [2]新疆大学新疆多语种信息技术实验室,新疆乌鲁木齐830046

出　　处：《计算机工程与设计》2019年第1期273-278,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61462083;61262060;61331011;61463048);国家973重点基础研究发展计划基金项目(2014CB340506);"自治区青年科技创新人才培养工程"青年博士基金项目(QN2015BS004)

摘　　要：为缓解目前维吾尔文机构名识别方法依赖于人工编写规则、识别效率低的问题,提出一种基于条件随机场模型(CRF)的维吾尔文机构名识别方法。根据维吾尔语的语言特性,结合词、词性、音节、机构名特征词表、地名词表等特征,实现维吾尔文机构名识别。实验结果表明,相比于基于规则的方法和隐马尔科夫模型(HMM),该方法不依赖于人工编写规则,识别的准确率和召回率较高。To alleviate the problems of heavily relying on manually written rules and low recognition efficiency in Uyghur organization name recognition,a conditional random field model(CRF)based Uyghur organization name recognition method was proposed.Uyghur organization names recognition was implemented,according to the linguistic characteristics of the Uyghur language,by combining the features of word,part of speech,syllable,the feature words of organization names,location names.Experimental results show that comparing to the rule-based method and hidden Markov model(HMM),the proposed method is independent of manually written rules,and high recognition accuracy and recall rates are achieved.

关 键 词：命名实体 机构名识别 维吾尔语 条件随机场 黏着语 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]