Links

Split Files

Split Files allows you to split each file in your project amongst your labelers. If you don't need each labeler to label the same portion of data this will be very useful. Using this feature, enables you to assign specific files to particular labelers. Read below for instructions on how to utilize this feature. Watch this 1:00 minute Loom video for visual reference.

Sample case

We have 2 txt files with 10k rows each. Our goal here is we want to divide these files equally to 2 labelers. Let's do the following!
  1. 1.
    Upload the files in a project as usual in Step 1.
  2. 2.
    In the Step 2, preview the files.
  3. 3.
    In the Step 3, set the task type and the label sets.
  4. 4.
    In the Step 4, choose two individuals as the labelers.
    1. 1.
      If the purpose in this project is only to speed up the labeling process and doesn't require the Reviewer judgement, we can set the consensus as 1.
    2. 2.
      If you still need the Reviewer judgement, set the desired consensus.
  5. 5.
    Select the Advanced button.
  6. 6.
    In the Advanced menu, go to Split Files section and change the dropdown into Number of Parts. Then, set whichever number makes sense for your project. In this example, we are going to set it to 2. Each file will be split into two parts. It is noticeable, you can see Part 1 and Part 2 as the suffix.
  7. 7.
    Select Distribute Automatically. Our system will distribute the files according to the consensus review that has been set previously. In this case, each file will be assigned to the two labelers I selected: Nadya and Demo.
    • Demo will get Harry Potter - Part 1 and Sherlock Holmes - Part 1
      • They will only see these file parts in his project
    • Nadya will get Harry Potter - Part 2 and Sherlock Holmes - Part 2
      • They will only see these file parts in her project

Additional information

  • File splitting occurs during project creation. You are unable to split or merge back the file parts after the project creation.
  • Labelers will only get the assigned parts.
  • Reviewer will get all the file parts.
  • Split files feature is applicable to span-based, row-based, bounding-box, and document-based projects.
Last modified 4mo ago