Textual Near-Duplicate Grouping via Non-Contiguous Character Sequence Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for identifying textual near-duplicates in documents are inefficient and prone to high false positives, requiring extensive processing resources and time for accurate word-by-word comparisons.

Innovation Solution

A method that processes documents by identifying non-contiguous character sequences and determining their counts, generating similarity metrics by comparing these counts between documents, and assigning documents to groups based on these metrics, reducing the need for full word-by-word comparisons.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If word-by-word comparison is performed to determine similarity between documents, then measurement precision is improved, but processing time increases

Engineering Contradiction:
Improvesimilarity determination accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the document comparison process into multiple stages: first using fast filtering techniques (character n-gram counts, word frequency comparisons) to eliminate clearly dissimilar documents, then applying more computationally intensive word-by-word comparison only to documents that pass the filtering stage. This hierarchical segmentation resolves the contradiction by achieving high precision through detailed comparison only when necessary, while maintaining fast initial screening for the majority of document pairs.

Inventive Principle:
Principle #1Segmentation

2Productivity

If heuristic techniques are used to quickly identify textual near-duplicates, then processing speed is improved, but false positive rate increases

Engineering Contradiction:
Improveprocessing speedVSAvoidfalse positive rate
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent introduces intermediate filtering techniques that act as mediators between fast heuristic methods and accurate word-by-word comparison. These intermediate filters (such as character n-gram overlap analysis and weighted word frequency comparisons) provide a middle ground that significantly reduces false positives compared to simple heuristic methods, while still being much faster than complete word-by-word comparison. This intermediary layer resolves the contradiction by maintaining high processing speed while substantially improving reliability.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Measurement precision

If extensive processing resources are allocated to accurate document comparison, then measurement precision is improved, but use of energy increases

Engineering Contradiction:
Improvesimilarity determination accuracyVSAvoidprocessing resources
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent segments the computational workload into hierarchical stages with increasing resource requirements. The majority of document pairs are processed using low-resource filtering techniques that consume minimal energy. Only a small fraction of document pairs that pass through multiple filtering stages are subjected to high-resource word-by-word comparison. This segmentation resolves the energy precision contradiction by allocating extensive processing resources only to the minority of cases where high precision is actually needed, while using minimal resources for the majority of comparisons.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS10242261B1System and method for textual near-duplicate grouping of documents
Publication Date: 2019.03.26 RELATIVITY ODA LLC
  • US10242261B1 patent drawing
  • US10242261B1 patent drawing
  • US10242261B1 patent drawing

AI summary

A method for efficiently grouping electronic documents that are likely textual near-duplicates includes processing first and second electronic documents to determine respective sets of character sequence counts. The processing includes, for each document, (1) identifying non-contiguous character sequences expressed within the document text, with each character sequence corresponding to a different starting position within the text and including at least a first character at the respective starting position and a second character at a pre-defined offset from the respective starting position, and (2) determining character sequence counts for each unique character sequence within the identified character sequences. The method also includes generating one or more similarity metrics, at least by comparing the sets of character sequence counts determined for the first and second electronic documents. The method may also include assigning, based on the similarity metric(s), the second electronic document to a same document group as the first electronic document.