Import documents

Upload documents

After successfully creating your project, you can upload your documents from the Build section.

1. Open your project.
2. Drag and drop the first batch of your sample documents in the Upload sample documents section.
   Tip: You can use the suggestions from the Recommendations sections. These suggestions will guide you through the process.
   The uploaded files are automatically processed (uploaded, digitized, classified, annotated).
3. Upload the next batch of your sample documents by clicking Upload.
   Tip: You can check the Recommendations sections for suggestions on what more is useful to upload. For example, if there are too few utility_bills documents, a suggestion is displayed:

   utility_bills has too few samples. Add at least 150 documents for an optimal data set size.

   There are two types of recommendations, one for classification and one for extraction models.

   Figure 1. Upload documents and recommendations
   

   

   
   
4. Review the uploaded documents.
   1. Expand the needed section (for example, invoices, receipts).
   2. Click on a document name.
   3. Check if the Document type filled in automatically is correct. You can change the document type using the drop-down list.

Every document type that you upload generates a prediction, along with a confidence score. However, some of these predictions could be false positives. To control this, you can set up a confidence threshold that only approves documents that surpass a certain level of confidence.

To identify the appropriate threshold, you need to run a series of test documents through your workflow. You can then record the results in an Excel spreadsheet to analyze which confidence threshold yields the most accurate results.

In order to apply a confidence threshold, you need to adjust the Minimum confidence parameter for the current activity you are using.

Import datasets

Once you have an exported dataset from a modern project, you can import it into a different project. For more information on how to export datasets, check the Export section from the Publish page.

1. Navigate to and open the project into which you want to import data.
2. Select Add document type and create a new custom document type.
   Note: You can only import zip datasets directly in a document type.

   Figure 2. Add document type
   

   

   
   
3. On the new custom document type, select Upload and choose the zip file of the classic project you exported. Wait for the upload to finish.
   Figure 3. Upload processing
   

   

   
   

Once the dataset is imported, the training start automatically.