A Natural Language Querying System Based on Semantic Parsing

• Main Authors: Hsin-Yi Kuo, 郭欣逸
• Other Authors: 呂瑞麟
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/n75sfu

碩士 === 國立中興大學 === 資訊管理學系所 === 106 === With the rapid development of artificial intelligence,everyone is looking for more convenient methods to interact with machine to make technology and information more accessible to daily life. All interaction between human and machines start by understanding the messages the human . However, it is not easy for machines to completely understand natural language expression of human, Natural Language Query therefore becomes a hot research field. In this work we build a Natural Language Query system based on DBpedia for answering English natural language query. We collected named entities in the DBpedia and establish named entities dictionaries to provide information for named entity mapping process to find the proper named entities for answer of the query , than we also indentify the DBpedia classes and properties from the query by mapping entity labels to verb and noun in the query .With entities founded in the query, we analyze grammar structure of the natural language query by using the syntax tree (Parsing Tree) ,we recursively analysis the node pattern in each level of syntax tree to understand the relation between entities ,and find out the answer. The experimental result proves that our method can answer the multi-tiered questions such as "Who are the parents of the wife of Juan Carlos I ? ". We applied our system to QALD-6 multilingual test question set in the QALD conference and compared it with other previous work. The experimental results show that our method achieves a higher recall than other approaches.