Search Query Parsing for Criterion-Based Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search methods are inefficient in providing accurate and complete search results, particularly for product and ecommerce searches, due to their probabilistic nature, which often excludes relevant sites and includes irrelevant ones, and struggle with specific criterion-based searches.
Innovation Solution
The system automatically converts keyword searches into real-time, context-specific, criterion-based searches by parsing search queries into fragments, determining categories, extracting filters, and customizing queries to target databases, ensuring more precise and relevant results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If probabilistic search algorithms are used to search large databases of websites, then the search system can process a wide variety of site types and quantities, but relevant sites are excluded or ranked at low levels while irrelevant sites are included and ranked highly
Solution Approach 1:
The patent segments the search process into distinct phases: (1) preliminary filtering of websites based on predefined criteria before the probabilistic search, and (2) post-search filtering of results. This segmentation allows the system to maintain versatility in searching diverse sites while improving precision through targeted filtering at critical stages.
Solution Approach 2:
The patent applies preliminary action by pre-filtering and categorizing websites before the actual search operation. Websites are pre-assessed against relevance criteria, and only qualifying sites are included in the searchable database, thereby preventing irrelevant sites from being ranked highly while preserving the ability to search diverse site types.
2Measurement precision
If the algorithmic selection and ranking process is made more restrictive to exclude less relevant sites, then fewer irrelevant sites are included, but a larger proportion of relevant sites are also excluded
Solution Approach 1:
The patent introduces an intermediary filtering layer that mediates between the probabilistic search algorithm and the final results. This intermediary applies specific criteria to retain relevant sites that might otherwise be excluded by restrictive algorithms, while simultaneously blocking irrelevant sites that restrictive algorithms would allow to pass through.
Solution Approach 2:
The patent dynamically adjusts search parameters and filtering thresholds based on the specific search query and context. This allows the system to optimize the balance between inclusivity and exclusivity for each search, preventing both the exclusion of relevant sites and the inclusion of irrelevant ones.
3Quantity of substance
If the algorithmic selection and ranking process is made less restrictive to include a greater proportion of relevant sites, then more relevant sites are included, but a significant proportion of less relevant sites are also included
Solution Approach 1:
The patent extracts and removes irrelevant sites from the search results through post-search filtering. The system first allows a broad search to capture all potentially relevant sites, then systematically extracts and eliminates irrelevant sites based on predefined criteria, thereby maintaining quantity of relevant sites while improving precision.
4Ease of operation
If rank-ordering methods are used to display search results, then the system can present results in an organized format, but the ordering appears random and bears no substantive relationship to the user's search
Solution Approach 1:
The patent implements dynamic rank-ordering that adapts to the specific search query and user context. Rather than using static or purely probabilistic ranking, the system dynamically adjusts the ordering criteria based on the search terms, site categories, and relevance factors, thereby maintaining organized presentation while ensuring the ordering reflects substantive relationships to the user's search.
Data Source
AI summary
A method for generating categorical and criterion-based search results from a search query including receiving the search query, generating one or more query fragments, determining a category corresponding to the search query, determining one or more filters applicable to the search query and one or more core search terms applicable to the search query based at least in part on the determined category and the one or more query fragments, generating at least one custom query for at least one target database in the one or more target databases based at least in part on the one or more filters, the one or more core search terms, the determined category, and one or more attributes of the at least one target database, and executing the at least one custom query on the at least one target database to generate a set of search results.