Before you begin using FORMS to process a particular form in production, it is essential that you take the time to test the form definition thoroughly. Your tests will reveal adjustments you can make that will improve interpretation. A high interpretation rate allows you to get the most out of FORMS and decrease the amount of manual verification.
This topic describes how to fine-tune interpretation manually.
While you are fine-tuning interpretation, scan forms only once. Export the images to files and use the image files to test interpretation. (Detailed instructions.) The reason for this is that scanners do not produce an identical result each time a form is scanned, and interpretation results can be affected. If you use the same image files throughout testing, you can be sure that the improvements you achieve are actual improvements and not a result of temporarily better scanning.
Specify that field-level statistics are to be saved. If you are looking at them, there is usually no need to compare the actual forms with the transfer file. In other words, you do not need to verify and transfer forms to see the interpretation result. After checking field-level statistics:
Delete the forms from the database.
Before each additional time you use the image files, use Windows Explorer to move them from the destination folder(s) back to their source folder.
Make your adjustments. (See suggestions.)
Reinterpret the forms. That is, import the image files again.
Use many files to test interpretation. Fine-tuning interpretation to a small sample of forms can actually make interpretation worse in production. A minimum of 10 forms is recommended, but more is better.
You can either make one change at a time to a single form definition or define several additional form definitions with different settings and validation routines to determine which work best for that particular form.
External interpretation engines
Tips for improving the detection of mark fields