Upgraded projects and classification
Any projects created in an earlier version of Kofax Transformation Toolkit are upgraded automatically to the new document set format and the new classifier format. Because there is only a single training document set for classification, you do not need to train each classifier separately.
When you first open an old project in Kofax Transformation Toolkit 6.3.0, your old separate training sets are upgraded and merged automatically. Each old training set is converted into a document subset within the classification training document set, and once you train your project in Kofax Transformation Toolkit 6.3.0, the old classifiers that are no longer needed are removed.
The option to determine whether a class is trained for layout classification or content classification is no longer tied to a document set. This is now an option in the Class Details. For each class, the "Train this class for layout classification" option is only selected if the old project had documents for this class trained with a layout classification view. Similarly, the "Train this class for content classification" option is only selected if the old project contains document for this class trained with a content classification view. If a class has no training documents, both training options are enabled for the class by default.
Because older projects could have a separate copy of the same document in several view training sets, the copy of a document that was used by the layout view never had recognition results even though the copy for a content classification view did have recognition results. In Kofax Transformation Toolkit, this is no longer the case. If a document is used for layout and content classification, only a single copy of the document exists in the Classification Training set. The same document is used for both classification methods. If a document is used for both layout classification and content classification, recognition is performed even though it is not required for layout classification, because it is required for content classification.