Acronym Detection via Token Similarity Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing text mining methods fail to effectively find acronyms and variants of acronyms in informal texts due to their reliance on exact string matching and manual user interaction, which is inefficient for texts with high typographical errors and varied abbreviation usage.
Innovation Solution
A text analysis method that breaks down texts and reference terms into tokens, generates candidates for mapping, compares characters using mapping rules, and calculates a confidence value to rank candidates, reducing the need for human interaction and improving variant detection in informal texts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If exact string matching is used to find acronyms, then precision is improved, but the method fails to detect variants in informal texts with typographical errors
Solution Approach 1:
The patent transforms the rigid exact matching parameter into a flexible similarity scoring system. Instead of requiring exact string equality, the system calculates similarity scores between candidate tokens and reference terms, allowing for typographical errors and variant forms while maintaining detection precision through threshold-based filtering.
Solution Approach 2:
The patent introduces dynamic adaptability by allowing the matching criteria to adjust based on the informal nature of the text. The system dynamically evaluates multiple candidate tokens against reference terms using flexible mapping rules that can accommodate variations in abbreviation styles, typos, and informal language patterns.
2Loss of information
If manual user interaction is required to examine text corpus and collect variants, then detection completeness is improved, but productivity deteriorates
Solution Approach 1:
The patent implements self-service by enabling the system to automatically detect, collect, and rank acronym variants without requiring manual user intervention. The automated process generates candidate tokens, compares them against reference terms using mapping rules, and produces ranked results that comprehensively identify variants while maintaining high productivity.
Solution Approach 2:
The patent replaces the mechanical manual examination process with an automated computational system. Instead of manual text analysis and variant collection, the system uses algorithmic token generation, automated similarity comparison, and computational ranking to achieve complete variant detection with high efficiency.
3Adaptability or versatility
If fuzzy string matching is allowed to detect variants, then adaptability is improved, but the number of misfits increases requiring more manual checking
Solution Approach 1:
The patent implements feedback through the confidence value mechanism. The system calculates confidence values for each candidate-token to reference-term mapping based on mapping rules and similarity scores. This feedback allows the system to automatically filter out low-confidence false positives while retaining high-confidence true matches, reducing the need for manual verification.
Solution Approach 2:
The patent applies partial action by selectively applying different levels of matching strictness. Instead of uniformly allowing all fuzzy matches, the system generates multiple candidate tokens and selectively evaluates them against reference terms using mapping rules, keeping only those that meet confidence thresholds, thus reducing false positives while maintaining adaptability.
Data Source
AI summary
A list of reference terms can be provided. Text and the list of reference terms can be broken down into tokens. At least one candidate can be generated in the text for mapping to at least one of the reference terms. Characters of the candidate can be compared to characters of the reference term according to one or more mapping rules. A confidence value of the mapping can be generated based on the comparison of characters. Candidates can be ranked according to their confidence value.