MetaServer > Help > Separate > Separate & Classify

With the Separate & Classify action, you can, using a pre-trained classification model to automatically classify documents in a set.

You can build your own Custom Classification model, using the Azure Document Intelligence Studio, to automatically classify and separate documents in a single step.

The resulting separation points and classification can be shown and altered in the Organizer‘s “Classification” mode.

To add a Separate & Classify action, select the action after which you want to insert the Separate & Classify action and press Add -> Separate & Classify. The Setup window will automatically open.

NOTE: If you want an operator to check the separation points and classification of each document, make sure the Organize action is set to be the next action.

TIP: The thumbnail on the right will follow you, so you can easily refer to the Setup window. Click on the thumbnail to zoom in.

First, add a description to your rule. Then, select a field that will hold the extracted document type.

01 – Resource and Endpoint: select your Azure AI Document Intelligence resource using the drop-down arrow. If you don’t see your resource listed, please make sure your resource was added in the Azure Resources setup, which can be found in the Admin Client’s Server tab.

Your endpoint will be automatically populated when you have selected your resource.

02 – Model: select your custom classification model using the drop-down arrow. If you don’t see your model listed, please make sure your model was successfully built in the Azure Document Intelligence Studio.

You can find a detailed guide on how to train and build a custom classification model here:
https://www.capturebites.com/metaserver/help/extract/120-150/#02-02-02-06

03 – Confidence level: for testing purposes, you can already apply the confidence levels using the the Preview function. You can force the a document to be labeled as orange (= less confident classification. This means that the separation and classification of this document needs to be verified in the Organizer) if it is below the specified, “Low” confidence percentage level.

If the confidence is lower than the “Very low” confidence percentage level, the document will be labeled as red (= low confidence classification. This means that the separation and classification of this document needs to be verified in the Organizer).

Only if the confidence level is higher than the “low” value, the document will be labeled as green (= very confident classification. This document doesn’t need to be verified in the Organizer).

Based on your tests, you can use the same confidence level settings in the Organizer action “Classification” mode’s setup.

In actual runtime, as soon as the operator manually confirms or corrects the separators + classification of a document, the separator will be labeled as green. Only when all separators are green, the user can “OK” all the documents of a set.

You can have a closer look at a page by double-clicking its thumbnail or pressing the Preview Page button to open the preview window. You can browse through the document by scrolling through the pages or clicking on them.

You can also see the total number of documents generated with your rules in the right corner of the status bar.