Text Similarity Quantification Using Shortest Operation Paths
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for evaluating text similarity, such as Edit Distance and Jaccard Distance, fail to consider both the minimum number of operations and visual similarity between text strings, leading to inaccurate assessments in applications like eKYC, where manual corrections to OCR texts require enhanced accuracy and reliability.
Innovation Solution
A device and method that quantify text similarity by obtaining shortest operation paths with edit pairs, determining similarity scores by summing historical similarity scores of these paths, and selecting a minimum score to evaluate the similarity between OCR and edit text strings, incorporating frequencies of operations from historical data to assess relevance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Edit Distance technique is used to evaluate text similarity, then the minimum number of operations is considered, but visual similarity between characters is ignored leading to inaccurate assessment
Solution Approach 1:
The patent merges Edit Distance technique with Jaccard Distance technique into a unified evaluation mechanism. The combined approach considers both the minimum number of operations (Edit Distance) and visual similarity between characters (Jaccard Distance), thereby improving text similarity evaluation accuracy while maintaining a manageable level of complexity through the integration of two established methods.
2Adaptability or versatility
If Jaccard Distance technique is used to evaluate text similarity, then visual similarity is ignored along with character order, but the evaluation is more comprehensive in terms of operations
Solution Approach 1:
The patent segments the text similarity evaluation process into distinct components: Edit Distance for operational transformation analysis and Jaccard Distance for visual similarity assessment. By dividing the evaluation into these separate but complementary segments, the system achieves both adaptability in handling different text scenarios and precision in measuring similarity through specialized metrics for each aspect.
3Productivity
If only minimum number of operations is considered for text similarity, then computational simplicity is maintained, but relevance between text strings is not quantified
Solution Approach 1:
The patent incorporates feedback mechanisms where the evaluation system not only calculates similarity scores based on Edit Distance and Jaccard Distance but also uses these scores to provide actionable feedback in eKYC verification processes. This feedback loop enhances reliability by allowing the system to make informed decisions about text corrections while maintaining processing efficiency through the use of established distance metrics.
Data Source
AI summary
Disclosed herein are computer-implemented methods; computer-implemented systems; and non-transitory, computer-readable media, for quantifying text similarity. One computer-implemented method includes obtaining a plurality of shortest operation paths including one or more edit pairs for correcting an optical correction recognition (OCR) text string with an edit text string, where each of the one or more edit pairs denotes an operation performable to a character of the OCR text string during correction by the edit text string. A plurality of similarity scores is determined, each corresponding to one of the plurality of shortest operation paths and determined by summing historical similarity scores of the one or more edit pairs of each of the plurality of shortest operation paths. A minimum one of the plurality of similarity scores is selected to quantify text similarity between the OCR text string and the edit text string.