Natural Language Query Translation via Keyword-Module Indexing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Natural language interfaces face challenges in translating queries into SQL when the database structure is unknown, particularly in non-relational databases or real-time memory databases, where data is only accessible through APIs or generated in real-time.
Innovation Solution
The system determines likely keywords in natural language queries and associates them with modules, using an indexer to convert source code text into keywords, and a results generator to rank responses based on module associations and context-free grammar, allowing for information retrieval without knowing the database structure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a semantic model with keyword associations is used to translate natural language queries into SQL, then query translation capability is improved, but the system becomes unable to handle databases with unknown or non-relational structures
Solution Approach 1:
The patent introduces an intermediary layer consisting of module associations and context-free grammar rules that mediate between natural language queries and data retrieval operations. Instead of directly mapping keywords to SQL queries requiring known database schemas, the system maps keywords to modules that can handle various data sources including those with unknown structures, thus resolving the contradiction between translation capability and adaptability
Solution Approach 2:
The semantic model is enhanced to associate keywords with multiple modules rather than specific database tables or columns. This universal association allows the same keyword-module mapping to work across different database types (relational, non-relational, real-time memory databases) and structures (known or unknown), making the system versatile while maintaining natural language query translation capability
2Measurement precision
If the system requires knowledge of database structure for semantic model associations, then query accuracy is improved, but the system cannot retrieve data from hidden or API-only databases
Solution Approach 1:
Modules serve as intermediaries between the semantic model and diverse data sources. The semantic model associates keywords with modules without requiring knowledge of the underlying database structure. These modules then handle the actual data retrieval from various sources including databases with hidden structures or API-only access, thus maintaining query accuracy while expanding database accessibility
Solution Approach 2:
The system segments the query processing into distinct components: keyword extraction, module association, and data retrieval. This segmentation allows the semantic model to focus on accurate keyword-module mapping for query understanding, while delegating structure-specific retrieval logic to individual modules, thereby maintaining accuracy without requiring global database structure knowledge
3Measurement precision
If keyword associations are manually adjusted to reflect database model changes, then query translation accuracy is improved, but the maintenance complexity increases
Solution Approach 1:
The system enables automatic updating of keyword-module associations when database models change. Instead of requiring manual adjustment of semantic model associations, the system self-adapts by detecting database model changes and automatically updating the appropriate module associations, thus maintaining query translation accuracy while reducing maintenance complexity
Solution Approach 2:
The system implements feedback mechanisms that monitor database model changes and automatically adjust keyword associations accordingly. When the underlying database model changes, the system receives feedback about these changes and automatically updates the semantic model associations to maintain accurate query translation, eliminating the need for manual intervention
Data Source
AI summary
Disclosed herein are a system, non-transitory computer-readable medium, and method for responding to natural language queries. Keywords likely to appear in a natural language query are determined and each likely keyword is associated with a module. A response to a natural language query comprises information generated by each module associated with a likely keyword appearing in the natural language query.