首页> 外国专利> SYSTEMS AND METHODS FOR WORD SEGMENTATION BASED ON A COMPETING NEURAL CHARACTER LANGUAGE MODEL

SYSTEMS AND METHODS FOR WORD SEGMENTATION BASED ON A COMPETING NEURAL CHARACTER LANGUAGE MODEL

机译:基于竞争神经字符语言模型的文字分割系统和方法

摘要

Execute a string algorithm on a title associated with a product to identify at least one product type associated with the product, use a machine learning algorithm to predict at least one product type associated with a product based on the title, during identification or prediction A system and method are provided for detecting inaccuracies in a product title, comprising detecting inaccuracies in a title based on at least one, and outputting a message to a remote device indicating that the title includes inaccuracies. Executing the string algorithm includes receiving a set of strings, generating a tree based on the received set of strings, receiving a title, and traversing the generated tree using the title to find a match. can do. Using a machine learning algorithm may include identifying words in a title, learning a vector representation for each character n-gram of each word, and summing each character n-gram.
机译:在与产品相关联的标题上执行字符串算法以识别与产品相关联的至少一种产品类型,使用机器学习算法在识别或预测期间,使用机器学习算法预测基于标题的产品相关联的产品类型,在识别或预测中 提供了用于检测产品标题中的不准确性的方法,包括基于至少一个检测标题中的不准确性,并将消息输出到指示标题包括不准确性的远程设备。 执行字符串算法包括接收一组字符串,基于所接收的一组字符串生成树,接收标题,并使用标题来遍历生成的树以查找匹配。 可以做。 使用机器学习算法可以包括标题中的识别单词,学习每个字的每个字符n-gram的矢量表示,并求和每个字符n-gram。

著录项

  • 公开/公告号KR102330819B1

    专利类型

  • 公开/公告日2021-12-01

    原文格式PDF

  • 申请/专利权人 쿠팡 주식회사;

    申请/专利号KR20200085559

  • 发明设计人 위 슈시;리 징;

    申请日2020-07-10

  • 分类号G06F40/279;G06F16/31;G06F16/35;G06F40/268;G06N3/08;

  • 国家 KR

  • 入库时间 2022-08-24 22:33:34

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