To improve the accuracy or efficiency of sessions, you can take advantage of the Pre-Classification Processing stage and the Token Identification process to run processes on each page only once even if you have multiple identification conditions or document classes. Token Identification is specifically designed to handle multiple conditions. For example, consider the following scenarios:
Multiple Identification Conditions
To efficiently use multiple identification conditions to assign documents to a particular class, you can configure processes that extract information from the documents in Pre-Classification Processing and configure the Token Identification process in First Page Identification to assign only documents that meet all the specified conditions to the document class.
Example: The Wonderland Police Department is scanning tickets, and wants to create a document class for citations for parking at an expired meter. All parking violations are printed on 297x420 millimeter cards, and they are the only documents that use paper this size. They configure a Page Size Identification process in Pre-Classification Processing to determine whether documents are on this size paper. They also configure an Optical Mark Recognition process in Pre-Classification Processing to determine which box indicating the type of violation is checked. They create a document class called "Expired Meter Parking Violations." In the First Page Identification stage for that document class, they configure a Token Identification process to assign the document to that class if both conditions are true: it is on 297x420 millimeter paper and the "Expired Meter" box is checked.
Multiple Document Classes
Pre-Classification Processing and Token Identification can greatly improve the efficiency or accuracy of sessions that involve multiple document classes. Rather than run all the identification processes for each class in turn, you can configure processes in Pre-Classification Processing and use Token Identification in each document class to sort the pages depending on the values stored in the tokens, which will improve the efficiency. Also, with Token Identification, you can create more accurate identifications.
Example: The City of Neverland Treasurer's Office is scanning invoices. They have affixed barcodes containing budget codes to each document, and will use the codes to sort them into 15 different document classes. They configure a Barcode process in Pre-Classification Processing to read each barcode once, and Token Identification in each of the 15 document classes to identify a document as belonging to the class if the token from the Barcode process contains the appropriate value. This will be much faster than if each document class contained its own barcode process, and each barcode had to be read multiple times until the identification succeeded.
Example: Nassau University is scanning undergraduate and graduate admission applications in a single session. The two types of applications use forms that are very similar, but not exactly identical. They configure Form Identification in Pre-Classification Processing with two master forms, one for each type of application. Each scanned application will be compared to both forms and assigned a token based on which it more closely resembles. When Token Identification is used in the document classes, it should be more accurate, since each application has been compared to both forms, whereas otherwise many forms might be mis-assigned to the first document class since they are so similar.