基于最大熵模型的中国人名自动识别  

Automatic recognition of Chinese name based on maximum entropy

作　　者：曹波[1] 苏一丹[1] 邓琦[1] 

机构地区：[1]广西大学计算机与电子信息学院,南宁530004

出　　处：《计算机工程与应用》2009年第4期227-228,234,共3页Computer Engineering and Applications

基　　金：国家自然科学基金No.60564001~~

摘　　要：用最大熵模型自动识别中国人名。首先对语料库的词性进行角色替换,然后用特征模板从角色替换后的语料库中提取出特征集,接着用IIS算法训练特征集的最大熵参数,最后用viterbi算法对初分词文本进行角色标注,并在角色序列的基础上进行模式最大匹配,从而实现中国人名的自动识别。在封闭测试实验中,识别准确率、召回率、F-值分别达到了85.4%、91.2%、88.2%。Authors use the maximum entropy model to recognize the Chinese name automatically.Firstly,authors replace the corpus＇s poses with roles,then,use feature template to extract feature set from the corpus which poses have been replaced with roles,thirdly,train the parameters of the feature set using IIS algorithm,finally,use the viterbi algorithm to tag the text which has been roughly segmented.The possible names are recognized after maximum pattern matching on the roles sequence.The closed test shows that the precision,the recall and the F-measure reach 75.6%,91.4%,82.8%.

关 键 词：中国人名识别 最大熵模型 VITERBI算法 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]