At present, Synapse is setup to recognise a few documents based on the text found in the OCR. You won't notice this because they're documents that I see frequently and so the words are hard coded into Synapse. Since it is working well, I plan to add the option of using your own classifier words.
