首页> 外国专利> COMPUTER PROGRAM, METHOD AND DEVICE FOR LEARNING HORN CLAUSE COMPOSED OF LITERAL OBTAINED BY EMBEDDING VARIABLE INTO NATURAL LANGUAGE

COMPUTER PROGRAM, METHOD AND DEVICE FOR LEARNING HORN CLAUSE COMPOSED OF LITERAL OBTAINED BY EMBEDDING VARIABLE INTO NATURAL LANGUAGE

机译:通过将变量嵌入自然语言而获得的构成文学的喇叭声的计算机程序,方法和装置

摘要

To mechanically learn and deduce a horn clause having a form which is close to a natural language.SOLUTION: A common partial character string in a single text being a learning source is replaced by using a variable which should be constrained by the same value. A horn clause and a co-occurrence tree using a literal which is obtained by embedding the variable into a natural language are obtained. A possibility of a notation fluctuation such as a variable name is automatically dissolved from an integrated view point of the variable name which is replaced from the same term, a character of a substituted value, and a positional relationship of the same name variable between a plurality of the single texts. The adoption/non-adoption and the reliability of the horn clause or the like are calculated by using an adaptable text number, and the adoption/non-adoption, the reliability and the necessity of the horn clause are decided depending on a degree at which the horn clause can derive a character string which does not exist in the text, or a degree allowing the derivation of an existing character string. A user is made to possess a sense by outputting the mechanically-learnt regularity from a text aggregation which is aggregated by being selected by the user. A single replacement prohibition character string which should not be replaced singly is used, and the user learns the horn clause by adding the texts which are conditionally merged to each other to the horn clause.SELECTED DRAWING: None
机译:为了机械地学习和推导具有接近自然语言形式的horn子句。解决方案:使用一个变量来替换单个文本中作为学习源的公共部分字符串,该变量应受相同的值约束。获得通过使用将文字嵌入自然语言而获得的文字的horn子句和共现树。从变量名的统一观点出发,诸如变量名之类的符号波动的可能性自动地被消除,该变量名由相同的术语代替,替换值的字符以及多个变量之间相同名称变量的位置关系单个文本。通过使用可适应的文本编号来计算horn子句等的采用/不采用和可靠性,并且horn子句等的采用/不采用,可靠性和必要性取决于其程度。 horn子句可以派生文本中不存在的字符串,也可以派生允许派生现有字符串的程度。通过从文本聚合输出机械学习的规律性来使用户具有感觉,该文本聚合通过用户的选择而聚合。使用了一个不应单独替换的禁止替换字符串,并且用户通过将有条件地彼此合并的文本添加到horn子句中来学习horn子句。

著录项

  • 公开/公告号JP2020047030A

    专利类型

  • 公开/公告日2020-03-26

    原文格式PDF

  • 申请/专利权人 NARERU SYSTEM CO LTD;

    申请/专利号JP20180175748

  • 发明设计人 NAKAMURA KEISUKE;

    申请日2018-09-20

  • 分类号G06N5/02;G06F16;G06F16/30;G06F40/151;G06F40/216;

  • 国家 JP

  • 入库时间 2022-08-21 11:37:13

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