首页> 外国专利> APPARATUS AND METHOD FOR ADAPTING LANGUAGE MODEL BASED ON DOCUMENT CLUSTERING

APPARATUS AND METHOD FOR ADAPTING LANGUAGE MODEL BASED ON DOCUMENT CLUSTERING

机译:基于文档聚类的适应语言模型的装置和方法

摘要

The present invention relates to a kind of device and a kind of method,For adjusting the language model of clustering documents. The device includes: grouped element to the entire corpus of documents group of cluster and at least one cluster, so that the similar field of document can be grouped for adjusting language model according to the method for the present invention; A kind of voice recognition unit executes speech recognition to generate for inputting voice, speech recognition result word grid; Cluster selector calculates distance value to center, while from the result and cluster of speech recognition, and selects at least one similitude clustering, is based on calculated distance value; One weight calculation unit, at least one the similar cluster for calculating weighting language model are based at least one similitude clustering of distance value; Language model interpolation unit generates interpolation language model and is based on weight and language model; Pass through interpolation language model to the word grid that the most note that resets scores is cut with the unit that resets scores. ;The 2014 of copyright KIPO submissions
机译:本发明涉及一种用于调整聚类文档语言模型的装置和方法。该装置包括:将元素分组到整个文档集的群集和至少一个群集的组中,以便根据本发明的方法可以将文档的相似字段进行分组以调整语言模型。一种语音识别单元执行语音识别以生成用于输入语音的语音识别结果词格;聚类选择器基于所计算的距离值,从语音识别的结果和聚类中计算到中心的距离值,并选择至少一个相似度聚类;一个权重计算单元,至少一个用于计算权重语言模型的相似聚类基于至少一个距离值的相似性聚类;语言模型插值单元基于权重和语言模型生成插值语言模型;通过插值语言模型传递到单词网格,该单词网格将重置分数的最注意部分用重置分数的单位剪切。 ; 2014年版权KIPO提交文件

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号