Branch Query Generation for Natural Language Search Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Natural language searches in structured databases often return irrelevant results due to lack of indexed metadata and inefficient querying, leading to increased system load and resource usage.
Innovation Solution
The system generates branch queries by categorizing search terms into entities such as business objects and conditions, using metadata to focus searches within specific database scopes, reducing unnecessary data processing and improving query accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If natural language search is used to query structured database, then ease of operation is improved, but search result relevance deteriorates due to lack of indexed metadata
Solution Approach 1:
The patent introduces an intermediary layer between natural language search and structured database. This intermediary parses natural language queries, maps them to structured query formats, and generates multiple branch queries that systematically explore different metadata fields and object types. The intermediary translates user-friendly natural language into precise database queries while maintaining search relevance through comprehensive metadata indexing.
2Quantity of substance
If natural language search queries all objects in structured database, then search result completeness is improved, but computing resource usage increases
Solution Approach 1:
The patent segments the search process into multiple branches, where each branch queries a specific subset of objects or metadata fields. Instead of querying all objects simultaneously, the system divides the search space into manageable segments (branches) and processes them systematically. This segmentation reduces computing resource usage while maintaining completeness by ensuring all relevant branches are explored.
Solution Approach 2:
The patent implements dynamic query generation where the search strategy adapts based on the query terms and identified objects. The system dynamically determines which branches to explore based on the natural language query analysis, adjusting the search scope and depth accordingly. This dynamic approach optimizes computing resource usage by avoiding unnecessary queries to irrelevant object types while ensuring complete search of relevant data.
3Device complexity
If metadata is not indexed in structured database, then device complexity is reduced, but search precision deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-indexing metadata fields before the search operation. The system prepares and organizes metadata information in advance, creating a structured reference framework that enables precise search queries. This preliminary organization of metadata allows the search system to quickly and accurately match queries against the correct fields without requiring complex real-time processing.
Data Source
AI summary
In one embodiment, a method receives a search query including terms formed in a natural language expression. The terms are included in entities that are categorized in types of entities. The method analyzes the entities for the terms to generate a set of branch queries that include different combinations of entities for the terms. The set of branch queries are ranked based on the types of entities associated with entities of the respective branch queries and a branch query is selected based on the ranking for the branch query. The method uses the entities of the branch query to perform a search of a structured database of structured objects.