Form Recognizer Solution Accelerator

Hi alexus98 , thanks for the comment. 

To answer your first question, I would recommend using a composite model. A composite model represents a collection of custom models and assigns them to a single model that encompasses your form types. During scoring, the composite model performs a classification step to determine which custom model the current form represents. You can then use the model ID of the composite model in this solution and let the Form Recognizer service determine which custom model to use. Check out more information here: https://docs.microsoft.com/en-us/azure/applied-ai-services/form-recognizer/compose-custom-models

To your second question, when you send document data to the REST API endpoint, an HTTP response is returned, which we capture in this solution as JSON. If you need to convert that JSON to a CSV file, you could attempt to do so as additional steps in the Analyze Results Logic App. There may be some complexity parsing the JSON for the specific fields you are interested in retaining in the CSV file. A quick search for “Logic App JSON to CSV” appears it is possible, but would require testing for your specific scenario. Check out more information here: https://stackoverflow.com/questions/48923628/logic-app-create-csv-table-from-json-output

Thanks!