Multi-data-source NL2SQL system based on semantic rules and multi-dimensional model

Abstract

The invention discloses a multi-data-source NL2SQL system based on semantic rules and a multi-dimensional model, comprising a service layer for realizing NL2SQL, the service layer comprising: a configuration module for configuring matching rules; a data model construction module used for constructing a data model capable of generating SQL statements by utilizing the table and field information ofthe database; and an intention recognition module used for analyzing and matching the natural statement according to the matching rule to obtain table and field information of a database, and generating an executable SQL statement by utilizing the data model. The system does not depend on annotation corpora and models, and NL2SQL based on intention can be achieved through simple configuration of an interface; multi-selection nesting is supported through the data model construction module, and flexible configuration can be achieved; the requirement for a data set is reduced through the intention recognition module, and the success rate of SQL generation is increased. Therefore, by adopting the system provided by the invention, the business scene can be quickly constructed according to different projects.

Description

technical field [0001] The invention relates to the technical field of intelligent search, in particular to a multi-data source NL2SQL system based on semantic rules and multi-dimensional models. Background technique [0002] In intelligent search, the process of computers understanding user query intentions has become a research hotspot in the industry. Before understanding user intentions, it is first necessary to convert natural language into an executable program language that computers can understand and generate accurate statement semantics. Natural Language to SQL (NL2SQL) is a method to convert the user's natural language statement into a computer-readable, executable, and semantic representation that conforms to computer rules. [0003] At present, in NL2SQL technology, professional knowledge reserves are needed for dictionary configuration, and there are certain requirements for nlp and model training; in addition, its requirements for data sets are relatively high...

AI Technical Summary

Problems solved by technology

[0003] At present, in NL2SQL technology, professional knowledge reserves are needed for dictionary configuration, and there are certain requirements for nlp and model training; (some natural languages ​​do not have corresponding SQL statements, the model should refuse to make predictions) for model training, the process is complicated and the completion time is long

[0004] Therefore, the existing NL2SQL technology is difficult to promote and apply in actual use

Images