Potential Layout Variations

The Potential Layout Variations feature helps users identify Structured layout variations that can be created to reduce the number of submitted pages that do not match to any layout variations in the system. This feature groups unmatched pages based on how likely they are to be the same page – these individual groups are called page groups.

In the Potential Layout Variations Viewer, you can view all the similar pages within a page group and use the merged image tool to create a new layout variation page for each group. Additionally, you can identify the other pages that likely comprise a multi-page layout variation by downloading a group's related pages. To learn more, see Download Related Pages.

Creating and running a Potential Layouts job

Note that Potential Layout Variation jobs must run in a separate trainer application. For more information, please see What is the Trainer?

To use the Potential Layout Variations feature, go to the Submissions > Potential Layout Variations, which displays all Potential Layout Variation jobs that have previously been run or are currently running. It will also display the Job ID, when the job was started, the job criteria, the number of pages included in the job, and the job status.

To run a new Potential Layout Variations job, follow these steps:

  1. Click on the Find Layouts button in the upper-right corner.

  2. Define the set of unmatched pages that the job should consider by providing a Submission Date range or Submission ID range.

  3. Click the Run button in the top right corner to start the job.

There are some important considerations when using this feature:

  • The minimum number of pages to run a job is 5 pages; the page groups smaller than this will not be included in a job and will remain marked as "Ungrouped".

  • A job can run up to 10,000 pages at once; if more than 10,000 pages have been selected, only the first 10,000 will be used.

  • The job may take several hours to run if a large set of pages is selected.

  • Only one Potential Layout Variation job can be run at a time – if a job is already running, the option to start a new job will be disabled until the current one completes.

Navigating the Potential Layout Variations Viewer 

To review a Potential Layout Variation job's results, access the Potential Layout Variations Viewer by clicking on the Job ID once the job is complete.

FindPotentialLayoutVariations.png

Header

To return to the list of all Jobs, click the ( X ) button in the upper right corner of the page.

Left panel - Job overview

The left panel gives an overview of the page groups produced by the job.

Job information can be shown by clicking on the i ) icon in the left panel. This information includes the status, when it was run, how long it took to complete, the criteria, and the number of submissions and pages.

This section also shows a graph of all pages included in the job, split by status:

  • Marked as Done - pages that have been designated as addressed by some user.

  • Not yet Addressed - pages have not yet been marked as done.

  • Not Grouped - pages that were included in the job, but not included in a group (likely because they were not sufficiently similar to other pages in the job).

To keep track of your work, click the Mark Pages as Done button next to the group name once you have either downloaded the merged image for future layout-variation creation or determined that the pages are not suitable for layout-variation creation.

When you review the page groups produced by the job, the goal should be to decrease the Not yet Addressed number to be as close to zero as reasonably possible. The page groups are listed in order of largest size, or how many pages there are within the group.

If there are any pages that have not been grouped by the machine, there will be a Not Grouped item at the bottom of the group list.

Right panel - Group details

The right panel shows all the pages that comprise a page group. You can click each individual image in the right panel to view it more closely in the middle viewer.

This area also includes a Merged tab where the machine attempts to separate content from the page's background. 

Reviewing job results

Merged tab

The toolbar above the page enables you to adjust the filter, undo & redo, preview, and zoom.

MergedTab.png

Follow these high-level steps when reviewing the job results:

  1. Review the images in each page group found by the job to determine if a layout variation should be created for the group.

  2. Next, determine if the grouped pages are from a single-page or multi-page document.

Manual Redaction

If the machine does not automatically remove all of the information present on the merged image, you can draw white boxes over the information to remove it manually. The objective here is to create the cleanest image because this will be the foundation of a new layout.

ManualRedaction.png

For single-page documents:

  1. Review the image shown in the Merged tab.

  2. You can use the Filter tool to adjust the contrast of the image as well as the amount of filled-in content on the page.

  3. If there is still personally identifiable information (PII) on the merged image, use the Manual Redaction feature to draw boxes over the remaining information.

  4. Once you are satisfied with the merged image, click Merged Image from the "Download" dropdown button to download the merged image that will serve as the foundation of a new layout.

  5. Finally, once you download a merged image for layout creation, or the group has been reviewed and found to not be suitable for a layout, use the Mark Pages as Done button to indicate that the page group has been reviewed and addressed.

For multi-page documents:

  1. For each page group, review the image shown in the Merged tab and identify which pages comprise your multi-page document.

  2. Follow Steps 2-5 from "For single-page documents". 

Optional step

Download Related Pages to see additional information about the page groups.

To do so, follow these steps:

  1. Click Download Related Pages from the "Download" dropdown button and extract the files from the downloaded ZIP file.

  2. Open the CSV which includes information about each related image. This information is useful in identifying which images are most likely a part of your multi-page document.

Creating layout variations from job results

Keep in mind that downloaded images may still contain PII or other sensitive information and may require manual redaction – review the images carefully prior to downloading, utilize the Manual Redaction feature if necessary, and ensure that any downloaded images are handled according to applicable data-retention policies.