Document Search Using Wildcard Likelihood and Context Scoring

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing document search methods face a trade-off between accuracy and comprehensiveness, with exact match searches lacking comprehensiveness and partial match searches lacking accuracy, particularly when dealing with typographical errors.

Innovation Solution

A document search device that computes the likelihood of hit character strings using a wildcard-based approach, incorporating forward and reverse likelihoods to balance accuracy and comprehensiveness, by replacing parts of the keyword with wildcards and evaluating the context before and after the hit character string.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If exact match search is used, then accuracy is improved, but comprehensiveness deteriorates

Engineering Contradiction:

ImproveaccuracyVSAvoidcomprehensiveness

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent changes the search parameter from exact character matching to likelihood-based matching using wildcards. By replacing exact match with wildcard-based partial match and computing likelihood scores, the system achieves both accuracy (through likelihood filtering) and comprehensiveness (through wildcard flexibility).

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent substitutes the mechanical exact matching mechanism with a probabilistic likelihood computation mechanism. Instead of rigid character-by-character comparison, the system uses likelihood computation that considers context before and after hit character strings, enabling flexible yet accurate search.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Adaptability or versatility

If partial match search is used, then comprehensiveness is improved, but accuracy deteriorates

Engineering Contradiction:

ImprovecomprehensivenessVSAvoidaccuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent introduces feedback through likelihood computation that evaluates context before and after hit character strings. The likelihood score serves as feedback to filter out inaccurate matches, allowing the system to maintain comprehensiveness while improving accuracy through contextual verification.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent applies local quality by evaluating the likelihood of character strings in their specific contextual positions. Instead of uniform matching, the system computes likelihoods based on local context (characters before and after the hit), allowing accurate discrimination between relevant and irrelevant matches.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS12602417B2Document search device, document search system, document search program, and document search method

Publication Date: 2026.04.14 RESONAC CORP

AI summary

A document search device includes an input reception unit configured to receive an input of a keyword of a document search, a document search unit configured to acquire, from a document, a hit character string matching a character string in which a portion of characters of the keyword is replaced with a wildcard, and character strings before and after the hit character string, and compute a likelihood of the hit character string, based on the hit character string, and the character strings before and after the hit character string, and a search result display unit configured to output a result of the document search based on the likelihood.