首页> 外国专利> REQUEST PARAPHRASING SYSTEM, REQUEST PARAPHRASING MODEL, TRAINING METHOD OF REQUEST DETERMINATION MODEL, AND DIALOG SYSTEM

REQUEST PARAPHRASING SYSTEM, REQUEST PARAPHRASING MODEL, TRAINING METHOD OF REQUEST DETERMINATION MODEL, AND DIALOG SYSTEM

机译:请求参数设置系统,请求参数设置模型,请求确定模型的训练方法和对话框系统

摘要

To allow a dialog system to flexibly cope even with various requests with different ways of expression.SOLUTION: A request paraphrasing system 120 comprises: a preprocessing part 130 which converts user input 56 into a word vector string; and a neural paraphrasing model 94 preliminarily learned by machine learning to receive the word vector string as input to paraphrase a request to be expressed by the word vector string to a request with higher probability that a question answering device 122 can answer than before paraphrasing. The request paraphrasing system may be constituted to determine whether or not the user input 56 is the request as preprocessing to paraphrase the request only when it is determined that the user input is the request. In addition, the request paraphrasing system may be constituted to classify to what request classification the input request belongs by a classification model 98 to be input in the neural paraphrasing model 94 as one of natures in paraphrasing.SELECTED DRAWING: Figure 1
机译:为了允许对话系统甚至以不同的表达方式灵活地应对各种请求。解决方案:请求释义系统120包括:预处理部分130,将用户输入56转换为单词向量字符串;以及通过机器学习预先学习的神经释义模型94,以接收词向量串作为输入,以将词向量串表示的请求释义为问题回答装置122可以比释义之前更可能回答的请求。可以将请求解释系统构成为仅在确定用户输入是请求时才将用户输入56是否是请求作为对请求进行解释的预处理。另外,请求释义系统可以被构成为通过分类模型98将输入请求所属的分类归入分类请求中,分类模型98将被输入到神经释义模型94中,作为释义中的一种性质。

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