Understanding Recognition Technology - An Insider's Guide (with eBook)1. © 2012 Parascript, LLC parascript.com
Understanding Recognition Technology
– An Insider’s Guide
Setting the Right Expectations for a Successful Implementation
2. 2© 2012 Parascript, LLC
eBook
Download the full ebook from Parascript
CLICK TO DOWNLOAD or go to http://hub.am/1lvNogg
Presentation
Slide Deck Available on SlideShare
QUESTIONS: email info@parascript.com
3. 3© 2012 Parascript, LLC
Agenda
Understanding the Recognition Process
Understanding Errors & Rejects
The Operating Point
Common Misconceptions
Context & Business Rules to Get the Most Benefits from Data
Why Results May Not Meet Expectations
Conclusion
4. 4© 2012 Parascript, LLC
Introduction
Understanding Recognition Technology – An Insider’s Guide, dives into the technical
aspects of how recognition software works, explains operational details, provides
examples, and identifies how and when human involvement is required in the process.
Companies, organizations and government entities worldwide use recognition technology to
process documents, forms, checks, mail and other important business content with high
speed and accuracy.
Types of Recognition Technology
Recognize machine print with OCR
(Optical Character Recognition)
Recognize handwriting, cursive handwriting, check marks, barcodes with ICR
(Intelligent Character Recognition)
5. 5© 2012 Parascript, LLC
Recognition Process
The typical recognition process:
input images are fed to the software.
After recognition, processed data is divided into two streams:
• accepted answers
• rejected answers.
After images are accepted, data includes:
• correct answers
• errors
6. 6© 2012 Parascript, LLC
TRADEOFF BETWEEN ERRORS AND REJECTS
Depending on application needs, users can either reduce rejects by
increasing error rates or reduce error rates by increasing reject rates.
Less Errors =
More Rejects =
More manual
processing
More Errors =
Less Rejects =
Less manual
processing
required
7. 7© 2012 Parascript, LLC
Reducing rejects by increasing error rates allows
companies to process more information
automatically, without the need for human
intervention. This lowers processing costs.
“ “
8. 8© 2012 Parascript, LLC
IMPORTANCE OF THE OPERATING POINT
Operating Point: a critical number since it is used to build a business case, establish an
ROI, and set the benchmark to measure against
Example
Operating Point is set to 85% read rate @ 1% error rate
100 Documents are processed
85 Documents will be read by the software properly with an error on 1
The remaining 15 documents are sent for manual processing
• For comparison, human (manual) processing typically yields a 3% error rate
The operating point is the most
critical number in recognition.“
“ The Rules For Using
Confidence Intervals
Download Understanding
Recognition Technology EBook for
the full story on the importance of
Operating Point
Page 8
9. 9© 2012 Parascript, LLC
COMMON MISCONCEPTIONS ABOUT RECOGNITION
Two Common Misconceptions about Recognition
all items in a stream have an equal level of difficulty,
regardless if they are processed automatically or
manually
that values should be recognized first and then use
rules and context
1
2
Common misconceptions relate to
the level of recognition difficulty
and the use of context and rules.
“ “
10. 10© 2012 Parascript, LLC
CONTEXT & BUSINESS RULES TO GET THE MOST BENEFIT FROM DATA
The effective use of context and business rules are tools that:
Allows users to process information faster
reduce reliance on manual data entry and related costs
derive the most benefits from their data
Context
In recognition, context explains the characteristics
or properties of data within a field
Rules
AKA “Logic”, rules apply “if/then” thinking or specific
scenarios to ensure that the data meets defined criteria
11. 11© 2012 Parascript, LLC
WHY RESULTS MAY NOT MEET EXPECTATIONS
Goal of Recognition
to produce the highest accuracy possible
Reality
type and quality of images being recognized can and often
produces significantly different results
individual characters, such as in handwriting, are more
ambiguous than others and harder to recognize
To Achieve Better Results
Better results from “poor quality” images and hard-to-read
characters, advanced handwriting recognition technology
employs the use of dynamic vocabularies
12. 12© 2012 Parascript, LLC
Advanced handwriting recognition technology
employs the use of dynamic vocabularies to produce
better results on poor quality images and hard-to-
read characters.
“ “
13. 13© 2012 Parascript, LLC
CONCLUSION
Companies and organizations with complex high volume processing requirements
benefit from automated recognition solutions.
Recognition technology has advanced substantially.
Current software options include sophisticated image analysis and pattern recognition
algorithms.
Businesses want to gain the most value for their software by achieving the highest
possible accuracy rates accompanied by lower costs.
Recognition Best Practices require that you understand:
Error and reject rates
confidence
operating point
context and business rules to set criteria to improve accuracy
14. 14© 2012 Parascript, LLC
Want to know more about recognition technology?
DOWNLOAD THE FULL WHITE PAPER
http://hub.am/1lvNogg
Check out Parascript solutions
FormXtra and FormXtra Capture.
Talk to an Expert
Click on Live Chat at the top of the Parascript website
www.parascript.com.
Understanding Recognition Technology – An Insider’s Guide