Complex multi-table SQL generation method and device based on bridging filling

Abstract

The invention relates to a complex multi-table SQL generation method and device based on bridge filling. A two-layer decoding architecture is adopted, and three parts are a semantic coding layer, an SQL template generation layer and an SQL detail filling layer respectively. The SQL template generation layer is a first decoding layer. The SQL detail filling layer is a first decoding layer, the SQL detail filling layer is a second decoding layer, in addition, the SQL generation model based on bridge filling only uses a sequence generation technology in the first decoding layer, the length of the SQL template is short, and compared with an independent sequence generation model, the calculation efficiency is greatly improved, and the calculation resource consumption is remarkably reduced.

Description

technical field [0001] The present application relates to the technical field of natural language processing, in particular to a method and device for generating complex multi-table SQL based on bridge filling. Background technique [0002] In the actual application scenario of SQL parsing, the natural text input by the user is generally a single table. For a single table, SQL statements can be generated by full matching. [0003] Compared with the single-table SQL parsing task, the multi-table SQL parsing task has a distinctive feature, that is, the database may contain multiple data tables, and there may be primary and foreign key relationships between the data tables. Compared with the single-table SQL parsing task, many The complexity of the table SQL parsing task is relatively high, which is mainly reflected in two aspects. First, as the number of data tables increases, the number of fields also increases, which may cause the input length to exceed the maximum length al...

AI Technical Summary

Problems solved by technology

[0003] Compared with the single-table SQL parsing task, the multi-table SQL parsing task has a distinctive feature, that is, the database may contain multiple data tables, and there may be primary and foreign key relationships between the data tables. Compared with the single-table SQL parsing task, many The complexity of the table SQL parsing task is relatively high, which is mainly reflected in two aspects. First, as the number of data tables increases, the number of fields also increases, which may cause the input length to exceed the maximum length allowed by the pre-training model; On the one hand, the SQL mode in the multi-table SQL parsing task is more complex, adding a variety of components (Having, Order by, Group by), set operations (Union, Except, Intersect), nested SQL, etc.

At present, there is no technical solution for automatic generation of SQL statements for multiple tables

Images