Natural Language Query to Nested Database Translation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Natural language bots struggle to effectively translate natural language queries into structured database queries in B2B environments due to variability, complexity, and ambiguity, leading to incomplete or incorrect results, especially when dealing with nested queries.
Innovation Solution
A system that utilizes a Natural Language Understanding (NLU) engine to identify intent and entities in natural language queries, determining valid database object references and creating either single-object or nested database queries by evaluating candidate query fields, operands, and contextual entities, and generating a query plan to accurately map user queries to structured database queries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If natural language processing is used to interface with B2B databases, then ease of operation is improved, but measurement precision and reliability deteriorate due to variability, complexity, and ambiguity in natural language
Solution Approach 1:
The patent introduces an intermediary natural language processing system that includes entity recognition, entity resolution, and query generation components. This intermediary layer translates ambiguous natural language queries into precise database queries by identifying entities, resolving their meanings based on context, and generating accurate SQL queries, thereby maintaining both ease of operation and measurement precision.
Solution Approach 2:
The patent segments the natural language query processing into distinct stages: entity recognition (identifying mentioned entities), entity resolution (determining what entities refer to in the database context), and query generation (constructing the database query). This segmentation allows each component to specialize in handling specific aspects of the translation process, improving overall accuracy while maintaining user-friendly input.
2Productivity
If natural language queries are translated to structured database queries, then productivity is improved, but device complexity increases due to need to handle multiple entity types and nested queries
Solution Approach 1:
The patent creates a universal entity recognition system that can identify multiple types of entities (people, organizations, locations, dates, monetary amounts) using a unified approach. The entity resolution component universally handles different entity types by determining what they refer to in the database context, and the query generation component universally constructs appropriate queries regardless of entity complexity, thereby managing productivity gains while controlling system complexity.
3Ease of operation
If implicit fields and fuzzy terms are handled in natural language queries, then ease of operation is improved, but manufacturing precision deteriorates due to semantic ambiguity
Solution Approach 1:
The patent implements feedback mechanisms in the entity resolution process where the system uses contextual information from the query and database schema to disambiguate fuzzy terms and implicit fields. The system feedbacks on the interpreted meaning to ensure accuracy, allowing users to formulate simple queries with implicit references while maintaining precise interpretation through contextual analysis and disambiguation.
Data Source
AI summary
The present disclosure describes a system, method, and computer program for converting natural language queries to structured database queries, including nested database queries. In response to receiving a natural language query for a database, an NLU model is applied to the query to identify an intent and entities associated with the query. The entities are tagged with an entity type that enables the system to identify any database object names, candidate query fields, operands, and contextual entities in the query. From the tagged entities, the system identifies one or more valid explicit, implicit, and indirect references to database objects in the user query. If there is only one valid reference to a database object in the user's query, the system proceeds with steps to create a single-object query. If there are valid references to two or more database objects in the query, the system proceeds with steps to create a nested database query. This includes grouping candidate query fields, operands, and contextual entities by independent object name, and evaluating each group separately to identify subject fields, conditional parameters, order/sort criteria, and record count limits for each group.