Automated Schema Annotation Generation for Natural Language Database Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating a database ontology for natural language interfaces to relational databases is a time-consuming and manually intensive process requiring deep knowledge of the database structure and schema, making it unsustainable for large or complex systems.
Innovation Solution
An automated system that extracts relational database metadata, prompts users for textual labels, and generates a schema annotation file to create a semantic model, allowing users to process natural language queries without expertise in the database structure or ontology, using machine learning to create schema-specific lexical rules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If manual schema annotation is performed by experts with deep knowledge of database structure and ontology, then the schema annotation file can be created with high precision, but the process becomes time-consuming and manually intensive
Solution Approach 1:
The system enables self-service by automatically generating schema annotation files through machine learning models that analyze database metadata and generate annotations without requiring manual expert intervention. The automated system serves itself by using the database structure information to create the schema annotation file independently.
Solution Approach 2:
The patent replaces the mechanical manual process of expert annotation with an automated machine learning system. The ML model substitutes human experts by analyzing database schemas and generating annotation files automatically, eliminating the need for manual knowledge-based annotation work.
2Reliability
If manual schema annotation is performed by experts with deep knowledge of database structure and ontology, then the schema annotation file structure and supported formats can be correctly implemented, but the process requires specialized knowledge and is not tenable for large or complex systems
Solution Approach 1:
The automated system performs self-service by using machine learning models to generate schema annotation files with correct structure and formats without requiring human experts. The system independently handles the complexity of database schemas and ontology relationships through automated analysis and generation processes.
Solution Approach 2:
The machine learning model acts as an intermediary between the database metadata and the schema annotation file. It mediates the complex transformation process by automatically interpreting database structure information and converting it into properly formatted schema annotations, eliminating the need for human experts to bridge this gap.
3Manufacturing precision
If traditional manual approaches are used to create schema annotation files, then users need deep knowledge of database structure, ontology, and schema annotation file structure, but this requirement makes the process inaccessible to users without specialized expertise
Solution Approach 1:
The system enables users without specialized knowledge to benefit from accurate schema annotations by using automated machine learning generation. Users don't need to manually create annotations; the system serves itself by generating accurate annotations automatically from database metadata, making the process accessible to anyone with database access.
Solution Approach 2:
The patent replaces the manual expert-driven annotation process with an automated machine learning system. This substitution eliminates the barrier of specialized knowledge requirements, as the ML model handles the complex analysis and generation tasks that previously required human experts with deep database and ontology knowledge.
Data Source
AI summary
Methods, systems and computer readable media are provided for automatically creating a semantic model of a relational database for processing natural language queries. A computing device automatically extracts relational database metadata. The computing device prompts a user to enter textual labels for columns of the extracted metadata. The computing device automatically generates a schema annotation file based upon the relational database metadata and the textual labels for the columns. A natural language query is processed for the relational database using the schema annotation file.