How to optimize Extraction Training folder
Issue
We are trying to lighten the weight of a couple of KTM projects as much as possible to be able to export them without problems. We have already eliminated the training part and compacted the databases, which has taken a lot of weight away from it. But we still have a folder that takes up quite a lot and that we really don't know if we need now that we have already compacted the databases and deleted the documents.
In the Extraction folder there are more than 3000 objects - documents in some cases from more than 5 years ago. Can we delete the contents of this folder or would it give us recognition problems?
Solution
In general, the Extraction folder contains trained data, image fingerprints, and images themselves. It is for the possibility of later re-editing. If you know now that you won't change the trained data that you have in the Extract folder, you can create a knowledge base file from it. Since it contains only the trained data and the fingerprints of the images, it provides the same extraction capacity, but without the images - the size of relevant KB file is significantly smaller. Once the KB file is created, you can add it to the project and remove the trained documents (ideal is to make a copy of that folder in case you need to add it to the project again and generate the KB file again for some reason):
Level of Complexity
Easy
Applies to
| Product | Version | Build | Environment | Hardware |
|---|---|---|---|---|
| Kofax Transformation Modules | All |