Velosyti

Tagging a question bank with AI

A question bank with a thousand questions and no tags is really just a very long document.

· 3 min read · For school principals and academic coordinators

Most schools already have a question bank — years of exam papers, worksheets and practice sets sitting in files. The trouble is finding what you need from it. A teacher preparing a revision test for "fractions, medium difficulty" ends up scrolling through old papers by memory, because nothing in the file is tagged by topic or level.

What tagging actually does

  • Reads each question and tags it by chapter, topic and sub-topic against your syllabus.
  • Estimates difficulty from how the question is phrased and what it asks the student to do.
  • Flags duplicate or near-duplicate questions that have crept in over the years.
  • Lets a teacher pull "20 medium-difficulty questions on fractions" as a list in seconds.

Signs your school needs this

If teachers keep a personal folder of "good questions" instead of using the school's bank, the bank has already failed as a shared resource. If setting a unit test still means one teacher manually picking questions from memory, the school is not using the years of material it already has.

What it costs you in effort

None of the source material changes — old papers, scanned or typed, go in as they are. Tagging thousands of questions against a syllabus takes AI a matter of hours, not weeks, though someone on staff should spot-check a sample before it goes live, the way any new system should be checked. After that, adding new questions each term is a small, ongoing job, not a repeat of the big one.

Questions to ask any vendor

  • Is tagging matched to your actual syllabus and textbook, or a generic curriculum that does not quite fit?
  • Can a teacher correct a wrong tag easily, and does the system remember that correction?
  • Does the tool keep your school's question bank private, or does it end up shared across other schools using the same vendor?
  • How long does a first pass over your existing question bank actually take, from files handed over to a searchable bank?

How we can help

Our reading model, Vizhi, already reads handwritten and printed answers for schools using Valarchi, our marking and reporting system for Tamil Nadu schools on the Samacheer Kalvi syllabus. Tagging a question bank uses the same reading ability, turned to a different job — organising what a school already has instead of grading what students wrote. See how Valarchi works on our Valarchi page. Tell us how big your question bank is and we will show you what a tagged version looks like.

References

  1. NCERT — National Council of Educational Research and Training
  2. UDISE+ — Unified District Information System for Education