Text Order Reversal for RTL Language Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current text extraction tools fail to correctly identify the logical order of text in documents, particularly those with right-to-left (RtL) languages like Arabic, when they are displayed in left-to-right (LtR) order, leading to reversed text that is unsuitable for analysis.

Innovation Solution

An apparatus and method that extract character data from documents, identify languages, determine the logical order, and generate confidence values to reverse the text if necessary, using language-specific rules and heuristics to correct the order for accurate analysis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If text extraction tools extract text in displayed left-to-right order from documents with right-to-left languages, then the extraction process is simple and fast, but the extracted text is in reversed order and unsuitable for text analysis

Engineering Contradiction:
Improvetext extraction speedVSAvoidtext logical order accuracy
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The system performs preliminary actions by extracting text in the displayed order first, then subsequently determining the logical order and reversing the text if needed. This preliminary extraction maintains speed while the later reversal step ensures accuracy for RTL languages.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system applies inversion by detecting when text is in displayed LtR order but should be in RtL logical order, then reversing the extracted text to correct the order. This inversion principle directly addresses the contradiction by flipping the text order when logical order differs from display order.

Inventive Principle:
Principle #13The other way round (Inversion)

2Measurement precision

If the system determines logical order and reverses text for right-to-left languages, then text analysis accuracy is improved, but the processing complexity increases

Engineering Contradiction:
Improvetext analysis accuracyVSAvoidprocessing system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system applies local quality by determining the logical order specifically for RTL languages rather than uniformly processing all text. The language identifier and order identifier components work locally on detected RTL text segments, adding complexity only where needed for accuracy.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The system changes parameters by detecting language type and logical order, then adjusting the text processing accordingly. When RTL language is detected, the system changes the processing parameter to reverse the text order, thereby improving accuracy without unnecessarily complicating the overall system.

Inventive Principle:
Principle #35Parameter changes

3Reliability

If the system uses language identification and logical order determination to reverse text, then the confidence in text order is improved, but the processing time increases

Engineering Contradiction:
Improvetext order confidenceVSAvoidtext processing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary language identification and logical order determination to establish confidence in the text order before proceeding with analysis. This preliminary assessment of language and order allows the system to make informed decisions about reversal needs while maintaining efficient processing flow.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2191396B1An apparatus for preparing a display document for analysis
Publication Date: 2013.05.29 INTERNATIONAL BUSINESS MACHINE CORPORATION
  • EP2191396B1 patent drawingFigure 1
  • EP2191396B1 patent drawingFigure 2
  • EP2191396B1 patent drawingFigure 3

AI summary

An apparatus for preparing a display document for analysis comprising: an extractor for extracting character data from the display document; an order identifier for determining a first order associated withprocessing of the character data and a second order associated witha logical order of the character data, and for determining whether the firstorder is different from the second order; and a reverse component, responsive to the order identifier determining that the first order is different from the second order, for reversing the character data.