首页> 外国专利> DIALOG ACT GRANTING MODEL LEARNING DEVICE, DIALOG ACT GRANTING DEVICE, METHOD, AND PROGRAM

DIALOG ACT GRANTING MODEL LEARNING DEVICE, DIALOG ACT GRANTING DEVICE, METHOD, AND PROGRAM

机译:对话行为授予模型学习设备,对话行为授予设备,方法和程序

摘要

PROBLEM TO BE SOLVED: To precisely grant a type of dialog act even to a speech including diverse and informal expression.SOLUTION: A speech analyzing part 11 performs morphemes analysis with a target speech sentence, and analyzes a category in a meaning system of each word. A feature amount extracting part 12 extracts, from a target speech sentence, character N-gram of various N which covers a specified percentage of the number of characters of a functional word as a feature amount. From a word train provided by abstracting a morphemes analysis result of the target speech sentence by using an analysis result of the category in a meaning system of each word, a word N-gram of various N that covers a predetermined percentage of the number of words in sentence end expression is extracted as a feature amount. A dialog act granting part 41 grants a type of dialog act to a target speech sentence by using a feature amount extracted from the target speech sentence and a dialog act granting model in which a type of dialog act learns, as learning data, a pair consisting of a type of each dialog act of a plurality of known learning speech sentences and a feature amount extracted from each of the plurality of learning speech sentences.
机译:解决的问题:甚至对包括多样化和非正式表达的语音都精确地赋予对话动作的类型。解决方案:语音分析部分11对目标语音句子进行语素分析,并分析每个单词的含义系统中的类别。特征量提取部分12从目标语音句子中提取各种N的字符N-gram,其覆盖功能词的字符数的指定百分比作为特征量。从通过使用每个单词的含义系统中的类别的分析结果来抽象目标语音句子的语素分析结果而提供的单词串中,各种N的单词N-gram覆盖了单词数量的预定百分比句末表达式中的“否”被提取为特征量。对话动作授予部41通过使用从目标语音语句中提取的特征量来将对话动作的类型授予目标语音语句,并且对话行为授予模型将对话行为的类型作为学习数据来学习一对对话行为。多个已知学习语音语句的每个对话动作的类型以及从多个学习语音语句中的每个提取的特征量。

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