Preview extracted data

Click See Extraction Results to review the extraction accuracy of the bot. This is the only way to see extraction results for table data, since the Designer does not show extracted table data during training. Alternatively, click Export to CSV to review the data in a CSV file.

The extraction result window displays the following information:
  • Successfully mapped fields and the extracted values.
  • Table headers and the values for each row.
  • Extraction errors, which are highlighted within a red box. Hover on the error to learn more, such as if there was a validation error or lack of OCR confidence.
Note:
  • The time required to display the extraction results can vary based on the complexity of the Python logic that you have created.
  • To ensure successful previewing of a training document, its entire end-to-end processing, including value extraction, execution of any Python logic (if applicable), and validation of all fields, must be completed within a 2-minute time frame.

    Training documents must be limited to a maximum of 20 to 30 pages. This ensures efficient processing and successful previewing within a 2-minute time frame.

Important: A training document can pass in spite of an error flagged by the validation pattern. Sometimes we want a validation pattern to flag an error, so it would not pass specific documents at a later stage in production; instead enter them in the manual validation queue.

Close the window to return to the Designer.

Important: If you upload 30 documents and all documents are classified into one group, you can see only one document for each unique layout.

Export data to a CSV file

You can export the extracted data to a CSV file for ease of review. When you click Export to CSV, the current document is downloaded to your device. To download other documents you must navigate to them using the < and > buttons.

The data in the downloaded CSV file is displayed exactly how it will appear when IQ Bot downloads data from a successfully processed document from a learning instance in production mode.