Natural Language Query Generation Without Data Warehouse Access
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems require manual generation of query instructions for data warehouses, which is time-consuming and prone to errors, and accessing sensitive data for training machine learning models to generate these instructions poses privacy risks.
Innovation Solution
A system using generative artificial intelligence and database query tools iteratively generates query instructions from natural language requests based on stored table descriptions, without accessing the data warehouse, by employing a generative AI model and database query tools to break down requests into step-by-step instructions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual generation of query instructions is used, then data privacy is preserved, but productivity is reduced and errors increase
Solution Approach 1:
The patent introduces an intermediary system consisting of a generative AI model and database query tools that mediates between the natural language request and the data warehouse. This intermediary automatically generates query instructions without requiring manual expertise, thereby improving both accuracy and productivity simultaneously
Solution Approach 2:
The patent replaces the manual mechanical process of query instruction generation with an automated system using generative AI and database query tools. This substitution eliminates the need for manual intervention while maintaining high accuracy through the structured interaction between the AI model and database tools
2Manufacturing precision
If sensitive data is accessed for training machine learning models, then manufacturing precision of query generation is improved, but harmful factors increase due to privacy risks
Solution Approach 1:
The patent extracts only the essential structural information (table descriptions, schemas, and formats) from the database, separating it from the sensitive actual data. The generative AI model is trained on this extracted structural information alone, achieving accurate query generation without exposure to harmful sensitive data
Solution Approach 2:
The patent introduces table descriptions as an intermediary layer between the sensitive data and the training process. This intermediary provides the necessary structural information for accurate query generation while acting as a barrier that prevents sensitive data from being accessed during model training
3Ease of operation
If manual generation of query instructions is used, then device complexity is reduced, but ease of operation deteriorates due to expertise requirements
Solution Approach 1:
The patent implements a self-service system where the generative AI model autonomously generates query instructions by interacting with database query tools. The system automatically retrieves table descriptions, formulates queries, and executes them without requiring user expertise, thereby improving ease of operation despite the underlying system complexity
Data Source
AI summary
Systems and methods are disclosed for converting natural language queries to a query instruction set for searching a data warehouse. To generate a query instruction set from a natural language query, a system iteratively uses a generative artificial intelligence (AI) model and database query tools to generate a query instruction set in a stepwise manner. The system and generative AI model do not require a priori knowledge of data table contents in the data warehouse, which may include sensitive information. In addition, the system does not require access to the data warehouse to generate the query instruction set. Instead, the system is implemented to use structure information from the data warehouse, including table lists (such as table names) and table format information (such as column names) of tables in the data warehouse, and the generative AI model is a generally trained model to generate the query instruction set.