首页> 外国专利> HISQA SYSTEMS AND METHODS FOR HUMAN INSPIRED SIMPLE QUESTION ANSWERING HISQA

HISQA SYSTEMS AND METHODS FOR HUMAN INSPIRED SIMPLE QUESTION ANSWERING HISQA

机译:用于人类启发的简单问题回答HISQA的HISQA系统和方法

摘要

The present specification describes a system and a method for determining how to automatically answer questions such as Where did Harry Potter go to school? Carefully built knowledge graphs provide rich sources of facts. However, a challenge still remains to answer factual questions in a natural language due to tremendous variety of ways a question is able to be raised. The present specification discloses embodiments of a system and a method for human-inspired simple question answering (HISQA), that is, a deep-neural-network-based method for automatic question answering using a knowledge graph. Inspired by human actions in such task, the embodiments first find a correct entity via entity linking, and then seek proper relation so as to answer the question, in which the above two steps are both implemented by deep gated recurrent networks and neural embedding mechanism.
机译:本说明书描述了一种用于确定如何自动回答诸如哈利·波特在哪里上学的问题的系统和方法。精心构建的知识图提供了丰富的事实来源。然而,由于能够提出问题的方式多种多样,用自然语言回答事实问题仍然是一个挑战。本说明书公开了用于人类启发的简单问题回答(HISQA)的系统和方法的实施例,HISQA是用于使用知识图自动回答问题的基于深度神经网络的方法。受到此类任务中人为行为的启发,实施例首先通过实体链接找到正确的实体,然后寻找适当的关系以回答问题,其中上述两个步骤均由深度门控递归网络和神经嵌入机制来实现。

著录项

  • 公开/公告号KR20170045135A

    专利类型

  • 公开/公告日2017-04-26

    原文格式PDF

  • 申请/专利权人 BAIDU USA LLC;

    申请/专利号KR20160133767

  • 发明设计人 LI LEI;XU WEI;DAI ZIHANG;

    申请日2016-10-14

  • 分类号G06F17/30;

  • 国家 KR

  • 入库时间 2022-08-21 13:27:38

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