Document Thumbnail Generation via Key-Value Extraction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Thumbnails and previews of documents are often too small to effectively convey important content, making it difficult for users to determine the relevance of the information without opening the document, thus causing inconvenience.

Innovation Solution

A method that analyzes document content, classifies key information, and generates detailed thumbnails or previews by consolidating key-value pairs, allowing users to view essential information without navigating through multiple pages.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of energy

If traditional small thumbnails are used, then bandwidth is reduced and download time is improved, but the ability to convey important content and enable content recognition deteriorates

Engineering Contradiction:
ImprovebandwidthVSAvoidcontent recognition
Core Design Contradiction:
Loss of energyVSLoss of information

Solution Approach 1:

The patent extracts and highlights only the most important text elements from the document, placing them in a thumbnail bar. This selective extraction allows the thumbnail to convey essential content information without requiring the full document to be loaded, thus reducing bandwidth while maintaining content recognition capability.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent transforms the traditional single-image thumbnail into a multi-dimensional representation that includes both visual thumbnail images and extracted text elements arranged in a horizontal bar. This dimensional expansion allows more information to be conveyed in a compact space, improving content recognition without proportionally increasing bandwidth requirements.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Device complexity

If client-side browser resizing is used, then no additional server resources are required, but visual quality deteriorates and bandwidth savings are lost

Engineering Contradiction:
Improveserver resourcesVSAvoidvisual quality
Core Design Contradiction:
Device complexityVSManufacturing precision

Solution Approach 1:

The patent performs text extraction, classification, and thumbnail generation on the server side before transmitting data to the client. This preliminary processing ensures high visual quality and accurate content representation is achieved upfront, avoiding the need for client-side resizing operations that would degrade quality.

Inventive Principle:
Principle #10Preliminary action

3Loss of information

If full document pages are displayed, then complete information is provided, but the interface becomes cluttered and usability deteriorates

Engineering Contradiction:
Improveinformation completenessVSAvoidusability
Core Design Contradiction:
Loss of informationVSEase of operation

Solution Approach 1:

The patent extracts only the most relevant text elements and displays them in a condensed thumbnail bar format, separating essential information from the full document. This extraction approach maintains information completeness for the displayed elements while dramatically improving usability by presenting a clean, organized interface.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent segments the document content into distinct thumbnail elements with clear visual separators. Each segment represents a discrete piece of information, making the interface less cluttered and easier to navigate while still providing access to complete information through the segmented presentation.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS11842143B1Techniques for thumbnail and preview generation based on document content
Publication Date: 2023.12.12 INTERNATIONAL BUSINESS MACHINE CORPORATION
  • US11842143B1 patent drawing
  • US11842143B1 patent drawing
  • US11842143B1 patent drawing

AI summary

A method, computer system, and a computer program product is provided for generating a detailed thumbnail and/or preview of a content. In one embodiment the technique comprises analyzing data obtained from a content and classifying it according to a plurality of specific types. A plurality of key information may be then extracted from the data according to the classification. A plurality of key-values are correlated to the plurality of key information so as to provide a plurality of key-value pairs. These pairs are consolidated accordingly to generate a thumbnail and/or a preview that at least renders the key information provided by the consolidated key-value pairs as a rendering.