一种新型的数据库自然语言查询实现方案
A New Implementation Scheme for"Database Natural Language Query"
富庭轩 1陈启明 1杨怀宇1
作者信息
- 1. 昆明理工大学 信息工程与自动化学院,云南 昆明 650504
- 折叠
摘要
当前深度学习技术对数据库自然语言查询的实现起到了很大的推动作用,但是仍存在难以实现复杂查询、准确率不高等问题.因此,提出了新型数据库自然语言查询实现方案,设计了一种基于谓词的、类Prolog语言的中间语言,以自然语言—中间语言—SQL的方式实现了很高的准确率,解决了自然语言与SQL语句之间语义差距大的问题,可以投入到实际应用中.还提出了一种学习样本自动生成的方法,降低了数据库自然语言查询技术的使用难度.
Abstract
The current Deep Learning technology has played a significant role in promoting the implementation of natural language query in databases,but there are still problems such as difficulty in implementing complex queries and low accuracy.Therefore,a new implementation scheme for natural language query in databases has been proposed,and an intermediate language based on predicate and language similar to Prolog has been designed to achieve high accuracy in a natural language—intermediate language—SQL manner,solving the problem of large semantic differences between natural language and SQL statements.It can be put into practical applications.A method for automatically generating learning samples has also been proposed,which reduces the difficulty of using natural language query techniques in databases.
关键词
数据库自然语言查询/NL2SQL/谓词推理/学习样本自动生成Key words
database natural language query/NL2SQL/predicate inference/automatic generation of learning sample引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据