Convert a New Samples database
A project from an earlier version of Kofax Transformation Modules that has online learning enabled, may still have training documents in the New Samples database. This is true even if you have fully imported and trained all New Samples before upgrading to the new version.
Contrary to the New Samples in the Extraction subset, the samples in the Problems subset are not removed from the database automatically. Any Problems documents marked as complete are filtered from the view automatically, but those documents are still present in the New Samples database.
If you upgrade to the new version, the best practice is to fully import and train all Extraction New Samples, as well as train and delete all documents in the Problems subset from the New Samples. Any remaining documents in the New Samples database are converted automatically in Knowledge Base Learning Server if a New Samples database is found during batch processing. This conversion is performed automatically for all projects that are synchronized to a batch class that has Knowledge Base Learning Server in its queue.
When you upgrade a project that is not synchronized to a batch class that runs Knowledge Base Learning Server during processing, you can convert the New Samples database manually in Project Builder. An empty New Samples document set is displayed when you open a project that has a New Samples database located in the online learning directory. The header of this document set shows that the New Samples database is not yet converted.
The conversion of the New Samples database can be very time consuming, especially if a large number of sample documents are remaining. During the conversion the New Samples database is locked. As a result, converting a large New Samples database can cause processing delays.
The manual New Samples database conversion is performed by following these steps:
- Open the Documents window if it is not already open.
- Right-click New Samples and select Convert New Samples Database.
OK to start the conversion.
Two document subsets are created, Problems and Extraction. These document subsets contain their respective training documents marked for online learning during Validation. If Classification Online Learning is enabled, the Classification document subset is also created and converted at this time.