Data Control Template

Hi BopBop,

The main issue here is that the default sets are conservative to prevent false positives.  To create some more targeted matches, creating your own template is definitely the way to go, and we encourage sharing these templates on this forum.

To create your template, you'll first want to fine-tune your requirements.  Do your files always contain <bank id><routing number><account number>?  If so, then you've got your fixed number of digits as the first thing you want to match.  After that, you'll also want to restrict your matches to acceptable subsets of those numbers... for the bank IDs, you'll probably want a list to choose from; for US routing numbers, you'll want the next set of numbers to be limited to a legitimate routing number (four different prefix variations followed by any 7-digit number).  Then for the account number, you'll need to limit your set to numbers that are legit account numbers (which, unfortunately, might be just about anything).

So, you'll probably be wanting something like (assuming # is a named number): <expression value="\b(?:722|###|###|###|###|###)(?: (?:<routing prefixes>)\d{7} )(?:\d{8,11})\b" count="1" weight="1" />

Remember that \b matches on a word ending, and so will be as happy matching '.' as an end of line or a space.

eg: 72208500001901258980 .4456 will match to the '.'  Removing the \b matches will enable you to match any numeric substring, which will increase FP risk while decreasing FN risk.

Now you have a few options: you can customize a rule for each style of document you send out, or you can attempt to generalize it to cover all cases.  The second option will be harder to debug if something goes wrong, and have more false positives.

If you don't want your templates to be available publicly, feel free to use the forum's message system to discuss your templates with other users privately.

If you want to share more, I'm sure others on the forum would be willing you provide more suggestions.

Hi BopBop,

The main issue here is that the default sets are conservative to prevent false positives.  To create some more targeted matches, creating your own template is definitely the way to go, and we encourage sharing these templates on this forum.

To create your template, you'll first want to fine-tune your requirements.  Do your files always contain <bank id><routing number><account number>?  If so, then you've got your fixed number of digits as the first thing you want to match.  After that, you'll also want to restrict your matches to acceptable subsets of those numbers... for the bank IDs, you'll probably want a list to choose from; for US routing numbers, you'll want the next set of numbers to be limited to a legitimate routing number (four different prefix variations followed by any 7-digit number).  Then for the account number, you'll need to limit your set to numbers that are legit account numbers (which, unfortunately, might be just about anything).

So, you'll probably be wanting something like (assuming # is a named number): <expression value="\b(?:722|###|###|###|###|###)(?: (?:<routing prefixes>)\d{7} )(?:\d{8,11})\b" count="1" weight="1" />

Remember that \b matches on a word ending, and so will be as happy matching '.' as an end of line or a space.

eg: 72208500001901258980 .4456 will match to the '.'  Removing the \b matches will enable you to match any numeric substring, which will increase FP risk while decreasing FN risk.

Now you have a few options: you can customize a rule for each style of document you send out, or you can attempt to generalize it to cover all cases.  The second option will be harder to debug if something goes wrong, and have more false positives.

If you don't want your templates to be available publicly, feel free to use the forum's message system to discuss your templates with other users privately.

If you want to share more, I'm sure others on the forum would be willing you provide more suggestions.