首页> 外国专利> QUESTION TRANSLATION SYSTEM USING DEPENDENCY PARSING TECHNIQUE AND SEMANTIC REPRESENTATION TECHNIQUE, AND METHOD THEREOF

QUESTION TRANSLATION SYSTEM USING DEPENDENCY PARSING TECHNIQUE AND SEMANTIC REPRESENTATION TECHNIQUE, AND METHOD THEREOF

机译:使用依赖解析技术和语义表示技术的问题翻译系统及其方法

摘要

The present invention relates to a question translation system using a dependency parsing technique and a semantic representation technique, to convert a question represented in natural language frequently used by people into an ontology query; and a method thereof. According to the present invention, the method comprises: a step (a) of constructing a grammar-meaning dictionary including a grammar-meaning, which is a pair formed with semantic parsing for a predetermined grammar pattern; a step (b) of understanding a grammatic relation of each element in a natural sentence of a natural query to form a tree structure, wherein the tree structure includes a node representing an element of the sentence and edges representing dependency between the elements; a step (c) of using a pattern of each node of the tree structure, and upper and lower nodes thereof as a grammar pattern to search for the corresponding grammar pattern in a grammar-meaning mapping dictionary so as to use searched meaning in node parsing; a step (d) of integrating a dependency parsing tree based on a combination rule, wherein the upper and lower nodes in the tree are set as first and second nodes, and semantic parsing of the first and second nodes are integrated when a second factor of a third factor in the semantic parsing of the first node has the same role as that of the second node; and a step (e) of generating a query when the meaning is integrated, wherein the query is generated by corresponding a first factor of integrated semantic parsing to a result of the query, corresponding a second factor to a structural condition of the query, and corresponding a fourth factor to a parameter filter of the query to combine the factors. Accordingly, a question represented in a natural language is converted into an ontology query, thereby being able to correctly extract knowledge represented in the natural sentence about the question.;COPYRIGHT KIPO 2018
机译:本发明涉及一种使用依存解析技术和语义表示技术将人们经常使用的自然语言表示的问题转换为本体查询的问题翻译系统。及其方法。根据本发明,该方法包括:步骤(a),构造包括语法含义的语法含义字典,该语法含义字典是通过对预定语法模式的语义解析而形成的一对。步骤(b),理解自然查询的自然句子中每个元素的语法关系,以形成树结构,其中,树结构包括表示句子元素的节点和表示元素之间依存关系的边;步骤(c):使用树结构的每个节点的模式及其上下节点作为语法模式,在语法含义映射字典中搜索对应的语法模式,以便在节点解析中使用搜索到的含义;步骤(d),基于组合规则对依赖关系解析树进行集成,其中,将树中的上,下节点设置为第一和第二节点,并在第二因素的影响下对第一和第二节点的语义解析进行集成。第一节点的语义解析中的第三因素与第二节点的作用相同。步骤(e),当所述含义被整合时生成查询,其中,通过将集成语义解析的第一因素与查询的结果相对应,将第二因素与查询的结构条件相对应来生成查询;以及将第四因子与查询的参数过滤器对应以组合这些因子。因此,将以自然语言表示的问题转换为本体查询,从而能够正确提取以自然句子表示的关于该问题的知识。; COPYRIGHT KIPO 2018

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