Transact

⌘K
  1. Home
  2. Transact
  3. Features and Functions
  4. Administrator Role and Fe...
  5. Batch Class Management
  6. Test Extraction

Test Extraction

Overview

Test Extraction enables you to test the extraction results without running an entire batch. By running Test Extraction, you can evaluate the result for a particular input and verify which kind of extraction is best suited for the particular document type.

Steps to Run Test Extraction

To begin the Test Extraction, we need to upload an image from the batch class screen either by dropping a file or clicking Upload Test Extraction File(s) located at the bottom of the screen. The uploaded image can be a single page or multipage TIFF or PDF file.

Figure 1: Upload Test Extraction File

Test Extraction can be run for different extraction plugins by altering two properties:

  1. Classification Types
  2. Extraction Plugins

Test Extraction requires the input document to undergo operations configured in the Document Assembler and KV Page Process plugin for the selected classification type. It supports three classification methods:

  • SearchClassification
  • BarcodeClassification
  • ImageClassification

On selecting ImageClassification in the Classification drop-down, the Document Assembler and KV Page Process plugins would perform only those operations that are necessary for Image Classification only, if the scripts are configured for both the modules it would be executed at the same time.

So to run the test extraction you need to configure all the relevant plugins in both the modules for the batch class.

Note: The extraction will run on the input document irrespective of whether the extraction switch value is ON or OFF. However, the plugins need to be configured in the batch class to perform the extraction and so does the Extraction Script plugin.

Functionality

On the test extraction screen, you will come across three buttons

  • Extract
  • Download
  • Clear

By clicking the Extract button, extraction will be performed on the input document and the result is generated. If no extraction result is returned, the “no values extracted” displays on the screen.

By clicking the download button, the extracted results XML file will be available for download. The schema of the downloaded XML file will be similar to batch.xml produced while running the batch. However, the batch has empty batch instance identifier fields.

By clicking Clear, the extraction results and the XML files will be cleared. This enables you to perform other extraction on the input document

Apart from these, you may click the close button to close the Test Extraction screen.

Result

After the extraction plugin returns a result it would be displayed as shown below. There are two ways to view test extraction results:

  1. Extracted DLF
  2. Data Table

Extracted DLF

Extracted Document Level Fields (DLF), also referred to as index fields, are the key-value extracted fields in a document type. The DLFs display index field details.

Using test extraction for Extracted DLF will display the following information:

  • Page ID
  • Page Name
  • Value Extracted
  • Field Details
  • Extraction Type
  • Confidence Level

The DLFs are displayed by their Field Order number.

Figure 2: Test Extraction Screen

The Format Values switch is for format conversion. Having this set to ON will apply the format conversion rule defined in the batch class.

When the Workflow switch is ON, test extraction (and classification) will use the settings of the batch class. In OFF state, you can use the options in the classification and extraction dropdowns to control the process. No other setting or configuration can be used in this state.

Data Table

In the case of table extraction, the data is populated under the DataTable section under the classified document. This will show test extraction data for all columns that have been configured. If columns have not been configured for table extraction, nothing will be extracted there.

If no value is extracted from the input document no Extracted DLF will appear.

Figure 3: Data Table Extraction Screen

Articles