Search Engine Query Weighting for Typo Tolerance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search engines face challenges in determining the relevance of data records due to vague and uncertain search queries, which can be exacerbated by typing errors and grammatical variations, leading to incorrect prioritization of results.
Innovation Solution
A computer-implemented method that processes search queries by determining a subset of similar or identical terms, applying a weighting distribution based on similarity measures, and calculating modified probabilities to assess the relevance of data records, thereby reducing the impact of typing errors and variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If search engines process vague and uncertain search queries directly, then processing speed is maintained, but search result accuracy deteriorates due to typing errors and grammatical variations
Solution Approach 1:
The patent applies preliminary action by pre-processing search queries to identify and correct typing errors and grammatical variations before the main search operation. The system analyzes the search query, detects potential errors, and generates corrected versions in advance, thereby improving search accuracy without significantly increasing processing complexity during the actual search execution.
Solution Approach 2:
The patent introduces an intermediary processing layer between the user's search query and the database search operation. This intermediary component analyzes the query, identifies typing errors and variations, and generates multiple corrected query versions that are then used to search the database, thereby mediating between the uncertain input and the need for accurate results.
2Adaptability or versatility
If search engines expand search terms to account for variations and errors, then search coverage is improved, but false positive results increase due to incorrectly spelled words occurring rarely in documents
Solution Approach 1:
The patent applies local quality by differentiating the treatment of different search terms based on their characteristics. The system identifies which terms are likely to be correctly spelled and which are likely to contain errors, then applies appropriate processing strategies to each term individually rather than uniformly expanding all terms, thereby improving reliability while maintaining adaptability.
Solution Approach 2:
The patent changes parameters such as term frequency thresholds and expansion criteria dynamically based on the specific search query and document corpus. By adjusting these parameters, the system optimizes the balance between search coverage and false positive reduction, ensuring that term expansion is applied appropriately without overwhelming the results with incorrect matches.
3Productivity
If search engines use simple term matching, then processing speed is maintained, but relevance ranking accuracy deteriorates due to vague search queries
Solution Approach 1:
The patent applies partial action by selectively applying advanced processing techniques only to specific portions of the search query that are identified as containing errors or variations. Rather than processing the entire query with complex algorithms, the system applies corrections and expansions only where needed, thereby maintaining overall processing speed while improving relevance ranking accuracy for the critical parts of the query.
Data Source
AI summary
The invention relates to a method implemented by a computer for searching in a plurality of data sets. In the method a search query is received and a partial quantity having terms similar to or identical to the search term is derived from a reference quantity. A similarity measure to the search term and the probability of the occurrence of the term is then determined for each term. Furthermore, a weighted distribution depending on the term is applied to the terms, and a modified probability is determined for the term. The data sets are further evaluated with respect to the relevance thereof to the search query, and at least one partial data set quantity is output as a function of the relevance value thereof. The invention further relates to a search engine for performing said method.