Character String Similarity Evaluation Using Position-Weighted Association Chains
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for comparing character strings are too imprecise, failing to account for swaps and other variations in character sequences, leading to inaccurate similarity evaluations.
Innovation Solution
A method that evaluates the similarity of character strings by determining associations within the strings, assigning numerical values to these associations, and forming an overall similarity measure based on chain relationships, with weights applied to positions to prioritize relevant agreements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional character string comparison methods are used, then the comparison process is simple, but the measurement precision of similarity evaluation is insufficient
Solution Approach 1:
The patent segments character strings into multiple positions and evaluates each position independently with position-specific weights. Instead of treating the entire string as a single unit, the method divides it into discrete positional segments (first position, second position, etc.), allowing precise control over the importance of each position in the similarity evaluation process.
Solution Approach 2:
The patent applies different evaluation criteria and weights to different positions within character strings. Each position can have its own weight factor, allowing the system to emphasize critical positions (like the beginning of a name) while de-emphasizing less important positions. This local differentiation enables more nuanced and accurate similarity measurements.
2Measurement precision
If position-weighted association chains are used to improve similarity evaluation, then the precision increases, but the computational complexity increases
Solution Approach 1:
The patent breaks down the complex task of string similarity evaluation into smaller, manageable segments by evaluating each character position independently. This segmentation allows the use of simple comparison operations at each position while achieving complex overall evaluation through the aggregation of position-specific results with appropriate weighting.
3Reliability
If swaps and variations in character sequences are considered, then the reliability of similarity detection improves, but the complexity of the comparison method increases
Solution Approach 1:
The patent introduces position weights as adjustable parameters that can be modified based on the specific comparison needs. By changing these parameters, the system can adapt to different scenarios such as detecting swaps, handling variations, or emphasizing certain positions, providing flexibility without requiring fundamentally different comparison algorithms.
Data Source
Figure 1
AI summary
The invention describes a computer-aided method for automatically valuating the similarity of two character strings which are stored in a computer or to which the computer has access via an interface. First of all, associations which are present in the character strings are sought using a specification stored in the computer. The sought associations are then valuated using a first rule stored in the computer, with cohesive associations - subsequently also referred to as association strings - being given a higher valuation for the similarity of the character strings than non-cohesive associations. Finally, a second rule stored in the computer is used to derive a value, particularly a numerical value, from the valuation of the sought associations as a measure of the similarity of the two character strings.