Generated Query Spaces for Accurate Structured Data Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semantic search technologies for structured data in cloud platforms face inefficiencies due to reliance on specific query phrases, leading to inaccurate results and increased resource consumption, especially when dealing with dynamic and structured documents.
Innovation Solution
The system generates a set of candidate natural language queries using large language models (LLMs) based on structured metadata, embedding these queries into vector spaces for comparison with user queries, allowing for more accurate and efficient retrieval of relevant documents.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If keyword-based search is used for structured data, then search implementation is simple, but search accuracy deteriorates due to loss of context
Solution Approach 1:
The patent introduces an intermediary layer between the user query and the structured data. This intermediary is a generated natural language query that translates the user's intent into a query format that can effectively search structured data, thereby improving search accuracy without requiring complex search algorithms to directly handle structured data formats.
Solution Approach 2:
The patent replaces traditional keyword-matching mechanisms with a semantic understanding mechanism. Instead of mechanically matching keywords against structured data fields, the system uses natural language processing to understand the semantic meaning of queries and generate appropriate search expressions, substituting the mechanical keyword-matching process with a more intelligent semantic approach.
2Measurement precision
If specific query phrases are required for semantic search, then search results become more precise, but user operation complexity increases
Solution Approach 1:
The system performs self-service by automatically generating the query phrases needed for searching structured data. Instead of requiring users to formulate specific query phrases, the system takes the user's high-level intent and automatically generates the necessary search queries, thereby maintaining search precision while significantly easing user operation.
Solution Approach 2:
The system performs preliminary action by pre-generating natural language queries based on the user's input and the structure of the data being searched. This preliminary query generation happens before the actual search execution, allowing the system to prepare optimized search expressions that will yield precise results without requiring users to manually construct complex queries.
3Productivity
If traditional search methods are used for dynamic structured documents, then system resource usage is reduced, but search efficiency deteriorates
Solution Approach 1:
The patent extracts only the necessary information from the structured documents to create a searchable representation. Instead of processing entire dynamic documents, the system extracts key attributes and relationships, creating a condensed search index that maintains the essential semantic meaning while significantly reducing the amount of data that needs to be processed during search operations.
Solution Approach 2:
The system performs preliminary action by pre-processing and indexing structured data into a searchable format before actual search queries are executed. This preliminary indexing organizes the data in a way that enables efficient retrieval, allowing fast search operations without requiring heavy resource consumption during the actual search execution.
Data Source
AI summary
Methods, systems, apparatuses, devices, and computer program products are described. An application server or a data processing system may generate a set of candidate natural language queries that correspond to a data object (e.g., document, report, assert) based on inputting a set of metadata associated with the data object into a large language model (LLM). The system may embed the candidate natural language queries into a first set of vectors, where a query space may include a collection of the first set of vectors related to the data object. In addition, the system may embed a natural language query received from a user into a second vector. The system may perform a vector-space comparison of the second vector to the first set of vectors or the query space, and retrieve a data object associated with the natural language query based on the comparison.