Fuzzy Matching Using Condensed Phonetic Keys

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing fuzzy matching technologies face challenges such as high computational complexity, costly brute-force methods, misleading results from strong matches on weak data, limited control over matching processes, and lack of modularity and extensibility, making them inadequate for all applications.

Innovation Solution

A computer system with key generators, a comparison engine, filter, analyzer, scorer, and pruner that pre-compute and store condensed keys for reference records, allowing for efficient fuzzy matching by reducing the candidate pool through statistical analysis and multiple stages of matching, enabling better control and modularity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If brute-force field by field fuzzy matching is performed against each individual record in a reference database, then comprehensive matching coverage is achieved, but computational complexity and processing cost become prohibitive

Engineering Contradiction:
Improvematching coverageVSAvoidcomputational complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments the fuzzy matching process into multiple stages: first generating candidate matches using phonetic algorithms (Soundex, Metaphone) on key fields, then filtering candidates through statistical analysis, and finally performing detailed fuzzy matching only on the reduced candidate set. This segmentation reduces computational complexity while maintaining comprehensive matching coverage.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary actions by pre-computing phonetic keys and statistical profiles for all reference database records before the actual fuzzy matching operation. This preliminary processing creates indexes and filters that dramatically reduce the number of records requiring full fuzzy matching, thereby reducing overall computational complexity while maintaining reliability.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If strong matching criteria are applied to identify potential matches, then match quality improves, but false positives increase when common names or weak data are involved

Engineering Contradiction:
Improvematch qualityVSAvoidfalse positive rate
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent dynamically adjusts matching parameters and thresholds based on statistical analysis of the data. Common names and fields with high occurrence frequencies are assigned different weightings and threshold requirements compared to rare or unique values. This parameter adjustment reduces false positives from common names while maintaining high match quality for genuine matches.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The system incorporates feedback mechanisms where match results and statistical analysis inform subsequent matching operations. The statistical profile of matched fields is continuously updated, allowing the system to learn from previous matches and adjust its criteria to reduce false positives while maintaining precision.

Inventive Principle:
Principle #23Feedback

3Reliability

If comprehensive fuzzy matching is performed without pre-filtering, then all potential matches are identified, but processing time and resource consumption increase significantly

Engineering Contradiction:
Improvematch completenessVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The matching process is segmented into distinct phases: candidate generation using phonetic algorithms, statistical filtering of candidates, and final fuzzy matching on the reduced set. This segmentation ensures that comprehensive matching is performed only on necessary records, maintaining match completeness while reducing processing time.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Phonetic keys and statistical profiles are computed in advance as preliminary actions, creating filters that rapidly eliminate non-matching records before the time-consuming fuzzy matching operation. This preliminary processing maintains match completeness by ensuring all genuine matches are captured while dramatically reducing overall processing time.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS7870151B2Fast accurate fuzzy matching
Publication Date: 2011.01.11 FAIR ISAAC & CO INC
  • US7870151B2 patent drawing
  • US7870151B2 patent drawing
  • US7870151B2 patent drawing

AI summary

A computer-implemented technique for fuzzy matching. This works quickly yet accurately to determine if a given computer-readable record is represented, by exact match or pretty close match, in a large collection of computer-readable records. Further tools may be provided to assess the character of the match.