Hybrid Summary Generation With Keyword Verification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing summary generation systems using machine-learned models are prone to inaccuracies due to data bias and false recognition of important words, leading to low accuracy in generated summaries.

Innovation Solution

A summary generation system that integrates a syntax-based summary and an AI-generated summary, with a determination process to verify the correctness of specific words in the AI summary against the original text, ensuring only accurate information is included in the final summary.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If a summary is generated using a machine-learned model, then the generation process is automated and efficient, but the accuracy of the generated summary is low due to data bias and false recognition

Engineering Contradiction:

Improveautomation of summary generationVSAvoidaccuracy of generated summary

Core Design Contradiction:

Extent of automationVSMeasurement precision

Solution Approach 1:

The patent introduces an intermediary verification process between the AI model and the final output. A determination unit acts as a mediator that checks whether specific words in the AI-generated summary actually exist in the original text, filtering out false recognitions while preserving automation

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system implements feedback by comparing the AI-generated summary against the original text to verify the presence of specific words. This feedback loop identifies and corrects errors where the AI model falsely recognizes or hallucinates words not present in the source material

Inventive Principle:

Principle #23Feedback

2Productivity

If only AI-generated summary is used, then the process is simple and fast, but the reliability of the summary is low

Engineering Contradiction:

Improvespeed of summary generationVSAvoidreliability of generated summary

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system performs preliminary verification of specific words before finalizing the summary. By checking whether key words from the original text are correctly included in the AI-generated summary, the system ensures reliability is established before output, maintaining speed through targeted rather than comprehensive verification

Inventive Principle:

Principle #10Preliminary action

3Ease of manufacture

If the AI model is trained with supervised learning data, then the model can be generated efficiently, but the summary accuracy is affected by the tendency of learning data

Engineering Contradiction:

Improveease of model generationVSAvoidsummary accuracy

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The determination unit serves as an intermediary that compensates for biases in the training data. By independently verifying whether specific words from the original text are correctly included in the summary, this mediator layer corrects systematic errors introduced during supervised learning without requiring retraining of the model

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS20250384201A1Summary generation system, summary generation method, and recording medium recording summary generation program

Publication Date: 2025.12.18 SHARP KK

AI summary

A summary generation system includes: an acquisition processing unit that acquires text data; a first summary generation processing unit that parses a text in the text data acquired by the acquisition processing unit and generates a first summary including a specific word; a second summary generation processing unit that generates a second summary of the text data using a summary generation model generated by machine learning; and an integration processing unit that integrates the first summary and the second summary and generates a summary corresponding to the text data.