Identifying papers

At some point the Plom system needs to know which paper belongs to which student and this can be done in several ways:

  1. Prenamed papers: Plom can produce papers with student names and IDs already printed on them.

  2. Automated ID reading — When tests are producing using Plom’s ID Template, the system can use machine learning to read the digits from the student-ID boxes and match against the classlist. In practice these appear to be over 95% accurate, but are not infallible.

  3. Manual association — The simplest method is for a human to just read the ID from the page and enter it into the system.

All these eventually require verification by a human.

Running the auto-identifier

This will only work if your test uses our idBox template because the code first looks for the outer bold rectangle and then tries to locate the 8 digits of the student number, based on known locations inside that larger box.

Currently, the django-server requires command line access to run the autoidentifier. TODO: xref here once those tools are in the docs Tracking issue for running this via the web UI: Issue #2990.

Running the auto-identifier (legacy server)

  1. Open the Legacy Manager tool, then “Progress” → “ID progress”.

  2. Optionally, adjust the top/bottom crop values, either manually or by clicking “Select interactively”. Only the vertical coordinate are used. The defaults probably work in most cases.

  3. Click “Recognize digits in IDs” which starts a background job. Click “Refresh” to update the output window.

  4. Click “Run matching tools”. This currently blocks and might take a few seconds (say 3 seconds for 1000 papers).

  5. Click “Refresh prediction list” to update the table view.

Caution

You should manually and carefully check the results (the Identifier client will show you these by default) because it does make mistakes, especially when there are additional names available in your classlist who did not write the test.

Manually identifying

This is typically quite quick compared to marking and you will not need to assign much person-time. Since it does not require any heavy thinking it can be a good task for:

  • the instructor-in-charge who is regularly interrupted by questions about papers,

  • a (reliable) marker who finishes their other tasks early, or

  • the scanner once they have finished scanning and uploading papers.

For now see https://plomgrading.org/docs/clientUse/identifying.html