Data String Comparison Using Typo-Pattern Walk-Forward Scanning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for comparing data strings, such as edit distance and bipartite comparison, are either inefficient due to processing entire string contents or require iterative processes, making them slow and unsuitable for fast, reliable similarity rating generation.
Innovation Solution
A method employing a left-to-right scan algorithm that preprocesses strings to eliminate insignificant differences, uses a Finite State Machine to compare strings, and scores similarity based on common typographical errors like insertion, deletion, substitution, and transposition, allowing for rapid comparison and intuitive scoring.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the edit distance method is used to compare data strings, then the comparison can be performed systematically, but the entire contents of both strings must be processed making it slow
Solution Approach 1:
The patent extracts only the most common error patterns (insertion, deletion, substitution, transposition) from the complete set of possible string differences. By focusing on these four specific error types that account for approximately 80% of typographical errors, the method processes only relevant comparisons rather than examining every character position systematically, thus reducing processing time while maintaining reliability for common errors.
Solution Approach 2:
The patent applies partial action by implementing a limited look-ahead mechanism that examines only a small number of character positions ahead when a mismatch is detected, rather than processing the entire string. This partial scanning approach processes enough information to identify common error patterns while avoiding the exhaustive processing required by traditional edit distance methods.
2Measurement precision
If the bipartite string comparison method is used, then excellent results are achieved in finding hidden words, but the method requires an enormously iterative process making it comparatively slow
Solution Approach 1:
The patent implements a walk-forward scanning algorithm that skips ahead through the string when matches are found or when error patterns are identified. Instead of performing exhaustive iterative comparisons at every character position, the method rushes through matching sections and only performs detailed analysis when mismatches occur, dramatically reducing the number of iterations required while maintaining the ability to detect hidden word patterns.
Solution Approach 2:
The patent performs preliminary processing by pre-identifying and eliminating spacing and punctuation differences before the main comparison process. This preliminary action prepares the strings in advance, allowing the subsequent scanning algorithm to focus only on meaningful character comparisons and reducing the iterative work required during the main comparison phase.
3Reliability
If a comprehensive string comparison method is used to handle all error types, then all typographical errors can be detected, but the processing becomes complex and slow
Solution Approach 1:
The patent applies local quality by implementing different comparison strategies for different situations: standard character-by-character comparison for matching characters, and targeted look-ahead patterns specifically for detecting insertions, deletions, substitutions, and transpositions. This localized approach applies the appropriate level of complexity only where needed, maintaining simplicity for the majority of comparisons while providing comprehensive error detection capability when required.
Data Source
AI summary
A comparison routine for comparing two data strings includes pre-processing, copying letters and digits into local buffers, comparing the contents of the local buffers using a left-to-right, walk-forward scanning algorithm, including selecting a first item from each buffer and comparing them and, if the two first items are the same, advancing to the next two items from each buffer, continuing with the comparison of subsequent buffer items until all pairs of items have been compared as long as no mismatch of items is found, and when a mismatch is encountered, the algorithm looks ahead in the data strings, searching for an identifiable pattern. A pair of independent search pointers are initially set to a first character in each data buffer and adjusted if a matching character pattern is found. A processing loop sequentially tests for a number of character patterns. If a matching character pattern is found, the match is scored, incrementing the score by one point for each matching character pattern that is found. If none of the character patterns is found, or the end of either string is reached, the processing loop is exited and the scoring data is collected.