Kofax Transformation Modules Invoice Pack 1.0 - Manual (Page 11)

Overview
Kofax Transformation Modules Invoice Pack – Getting Started Guide 11
However, in the case of Invoice Number and Purchase Order Number, if an
invoice has been matched to a supplier, it may be possible to retrieve a regular
expression from the supplier database that has been written expressly for that
supplier’s invoices. If a customized regular expression can be found, any
results that match this regular expression are passed through with a high
confidence. If no customized expression is present, a generic regular expression
is used, and results passed on for further logic as with other fields.
Format Locators
Once the results have been retrieved by the regular expression locators,
keywords are used to pick out possibilities. These keywords have been
generated based on an analysis of a large numbers of invoices for a given
locale. The Invoice Pack projects can be extended further to include additional
keywords as and when required.
Special Locators
For some fields, special locators exist that perform additional logic. In
particular, the Invoice Header Locator enables high-performance extraction of
subtotals, tax amounts and totals based on the relationship between these
values. Where available, this functionality is used.
The results of the various methods above are combined together using
standard evaluators, based either on the highest confidence result, or in some
cases, the first available result. This decision is based on the type of locator
used – some locators are more reliable than others.
After extraction, the results are copied into the invoice fields.
Additional logic then determines the OCR confidence of the best result. If the
confidence is below a user defined percentage threshold (see configuration
section of this document), the result is marked as invalid.
Validation
The Invoice Pack ships with a pre-defined validation form.
This validation form includes standard rules to determine the validity of a
result. Dates and amounts must be formatted correctly, required fields must
contain values, and amount fields must add up correctly.