Natural Language Database Search with Semantic Query Expansion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional database searching requires users to have prior knowledge of the target subject matter, leading to inefficient and user-unfriendly experiences due to the need for specific keywords and multiple rounds of query refinement.
Innovation Solution
A system utilizing a machine learning model to process natural language queries, determine semantically similar queries, and search databases to provide relevant information, including citations, without requiring prior knowledge.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional database searching methods are used, then search accuracy can be maintained with precise keywords, but user accessibility deteriorates due to requiring prior domain knowledge
Solution Approach 1:
The patent introduces an intermediary system comprising a language model and search query optimization module that translates between natural language and database search queries. This intermediary automatically generates and refines search queries based on user input, eliminating the need for users to possess domain knowledge while maintaining search accuracy through automated query optimization and expansion techniques.
2Measurement precision
If multiple rounds of query refinement are performed to improve search results, then search accuracy improves, but time consumption increases
Solution Approach 1:
The system performs preliminary actions by automatically generating multiple refined search queries and executing them in sequence without user intervention. The search query optimization module proactively creates expanded queries, retrieves results, and refines subsequent queries based on initial results, completing what would traditionally require multiple manual refinement rounds in a single automated operation, thereby reducing time consumption while maintaining accuracy.
Solution Approach 2:
The patent implements continuous useful action through automated iterative query refinement where the system continuously generates improved queries based on previous results without idle periods. The search process flows continuously from initial query generation through multiple refinement cycles to final results, eliminating the discontinuous manual interaction required in traditional approaches and maintaining high search accuracy throughout the continuous process.
3Ease of operation
If traditional search methods are used, then database querying capability is maintained, but user friendliness deteriorates due to requiring specific search syntax and keywords
Solution Approach 1:
The patent implements universality by creating a multi-functional system that handles both natural language processing and structured database querying through a single integrated interface. The language model and query optimization module serve multiple functions: translating natural language to search queries, expanding queries with relevant terms, refining queries based on results, and adapting to different database types, thereby maintaining full database querying capability while dramatically improving user friendliness.
Data Source
AI summary
Systems and methods of searching content using one or more machine learning models are disclosed. In response to a user query request, a user query is received at a first user interface. One or more search terms are determined using a query parser based on the user query. One or more semantically similar queries are determined based on the one or more search terms. In accordance with a determination that the user query and the one or more semantically similar queries have not been previously searched, search for information related to the one or more search terms from one or more databases using the one or more machine learning models. An output to the user query is generated based on the information. The output includes at least a portion of the information and one or more citations associated with the information. The output is displayed at the first user interface.