Is it possible to use one Custom Document Extractor for different documents? It would be provided training and testing data for each document, but would this affect the overall efficiency of the processor? Is it recommended to create a processor per document?
It is recommended to create a separate Custom Document Extractor for each document type you want to process. You can try to use the same processor for multiple types if they have similar structures, but generally the extraction quality will work better the more specific the training data the processor is trained on.
You can also create a Custom Document Classifier to classify unknown document types, then use that output to send each document to the appropriate extractor processor.