Hidden Cycle Evidence Booster for Question Answering

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Question answering systems often fail to discover less-apparent patterns of evidence that span across multiple documents, as they primarily focus on apparent patterns within individual documents, leading to missed insights.

Innovation Solution

An information handling system converts source evidence into a frequency-based representation using techniques like discrete Fourier transform, identifying hidden cycles and extracting hidden evidence to process requests effectively.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If question answering systems focus on apparent patterns within individual documents, then the search and question answering process is simple and efficient, but less-apparent patterns of evidence that span across corpora remain undiscovered

Engineering Contradiction:

Improveloss of hidden evidence patternsVSAvoidcomplexity of evidence analysis system

Core Design Contradiction:

Loss of informationVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary component (hidden cycle evidence booster) that transforms source evidence into frequency-based representations and identifies hidden cycles. This intermediary processes the raw evidence data to reveal patterns that are not directly observable, thereby reducing information loss without requiring the entire question answering system to become significantly more complex.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system changes the parameter representation of evidence from raw text form to frequency-based representation through mathematical transformation. By converting evidence into frequency domains and identifying cyclic patterns, the system reveals hidden information that was not apparent in the original form, thus reducing information loss while maintaining manageable system complexity.

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If question answering systems use domain-specific corpora with specific vocabularies, then accuracy for domain questions is improved, but patterns spanning across different domains may be missed

Engineering Contradiction:

Improveprecision of domain-specific answersVSAvoidability to discover cross-domain patterns

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The hidden cycle evidence booster serves a universal function that can be applied across different domain-specific corpora. The frequency-based representation and hidden cycle detection methods are domain-agnostic, allowing the system to maintain precision within specific domains while also being adaptable to discover patterns that span across multiple domains and vocabularies.

Inventive Principle:

Principle #6Universality (Multi-functionality)

3Quantity of substance

If the system extracts all evidence from source documents, then complete information is available, but the system cannot distinguish between apparent and hidden patterns

Engineering Contradiction:

Improvequantity of extracted evidenceVSAvoiddifficulty of detecting hidden patterns

Core Design Contradiction:

Quantity of substanceVSDifficulty of detecting and measuring

Solution Approach 1:

The patent replaces direct textual analysis with a mathematical transformation approach. By substituting the mechanical process of pattern recognition with frequency-based mathematical analysis (Fourier transform and spectral analysis), the system can detect hidden cyclic patterns that are difficult to identify through conventional text analysis methods, thus reducing the difficulty of detecting hidden patterns while maintaining complete evidence extraction.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach enables the discovery of hidden patterns and cycles, improving the accuracy of question answering by uncovering evidence that may not be immediately apparent, leading to better hypothesis generation and forecasting capabilities.

Implementation Method 1

an information handling system converts source evidence extracted from a set of documents to a frequency-based representation of the source evidence

Methodology Applied

Scientific EffectDiscrete Fourier transform:

Data Source

PatentUS10579728B2Hidden cycle evidence booster

Publication Date: 2020.03.03 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

An approach is provided in which an information handing system converts source evidence extracted from a set of documents to a frequency-based representation of the source evidence. The frequency-based representation includes multiple signals that each corresponds to an evidence type in the source evidence. The information handing system selects one of the signals that indicates a hidden cycle corresponding to a frequency at which one of the evidence types occurs in the source evidence and extracts hidden evidence from the source evidence based on the hidden cycle to process a request.