An intelligent number retrieval system and method based on natural language understanding

Abstract

The invention discloses an intelligent data retrieval system and method based on natural language understanding, including a user natural language interaction module, Search Engine, Schema Graph, Tokenizer, Tree Builder, and SQL Generation, and the user natural language interaction module transmits natural language to the outside The tokenizer performs external segmentation to separate words, and the Tokenizer recognizes the meaning of each separated word and combines words into phrases, and finally selects the best combination and interpretation automatically or interactively with the user; Tree Builder builds a Query Tree based on the node sequence in the Tokenizer, SQL Generation converts Query Tree into SQL into the database DB, and feeds it back to the user's natural language interaction module for user reference. It can support more question forms and richer complex queries, and can use less information without Achieving higher accuracy on the MAS dataset with user interaction.

Description

technical field [0001] The invention relates to the field of computer science, in particular to an intelligent data retrieval system and method based on natural language understanding. Background technique [0002] The natural language interface of the database provides a simpler and more habitual way for people to access the database. Even people who lack computer expertise can easily obtain the data in the database by using natural language query through this interface. . Using such a query method, users do not need to master complex structured query languages ​​(such as SQL), nor do they need to understand the table structure of the database. However, the construction of natural language interfaces for databases has always been a difficult problem. Currently, the most mainstream solution is NaLIR , ATHENA, etc. [0003] figure 1 The system architecture of NaLIR is shown. The whole system consists of three main parts: question parsing part, interactive communicator and ...

AI Technical Summary

Problems solved by technology

[0011] Aiming at the deficiencies in the prior art, the purpose of the present invention is to provide an intelligent number retrieval system and method based on natural language understanding, which can support more question forms and more complex queries, and use less information Can achieve 89% accuracy on the MAS dataset without user interaction (same as NaLIR, ATHENA)

Images