What's new in using OCR output in a Citizen Science Workflow