PROPOSED PROJECT
Class Understanding Check
Show teachers which ideas students have misunderstood in written quiz answers. Compare answers with the marking guide, group common mistakes, and suggest a follow-up exercise for the teacher to review, with the original answers beside each finding.
The same wrong answer can hide different thinking.
24 fictional responses
Question / 1/3 + 1/4
“1 + 1 = 2; 3 + 4 = 7; answer 2/7”
The response explicitly adds both numerators and denominators.
“2/7”
The result suggests the same mistake, but the teacher needs to check the reasoning.
See the pattern.
Keep the original answer.
Twenty-four fictional responses answer 1/3 + 1/4. Twelve use the correct result, six show or suggest adding the denominators, four keep numerators unchanged when making twelfths, and two are blank. The review shows source answers S-007 and S-014 beside the proposed denominator-error group. These are suggested response categories for a teacher, not student diagnoses or automatic grades.
WHAT IS USUALLY DONE
A teacher reads quiz answers one at a time and makes notes about recurring mistakes. A score alone does not reveal whether learners added denominators, misunderstood equivalent fractions, or left the question blank. Different reasons can be combined into one broad “needs help” group.
WHAT SOVRUN CAN DO FOR YOU
Sovrun would build a review screen that places each written answer beside the marking guide and suggests groups of similar reasoning. Teachers can inspect representative answers, move misclassified responses, and distinguish an explicit error from an answer with too little explanation. The tool drafts a follow-up task for the specific idea the teacher wants to revisit, with no automatic change to grades.
Choose the follow-up
for the specific idea.
Six responses are provisionally grouped around adding denominators. Four suggest a different error when rewriting equivalent fractions. The teacher can inspect each answer and move it before choosing the next task.
Correct result
Check that the explanation supports the correct result; accept valid alternative methods.
S-001 · S-002 · S-003 · S-004 · S-005 · S-006 · S-010 · S-011 · S-012 · S-013 · S-017 · S-019Added denominators
Use equal-sized parts to explain why adding denominators does not preserve the quantities.
S-007 · S-008 · S-014 · S-016 · S-020 · S-022Equivalent-fraction error
Revisit equivalent fractions: when the denominator changes, the numerator must change consistently.
S-009 · S-015 · S-018 · S-021Blank response
Ask what prevented a response. A blank alone does not show the learner’s reasoning.
S-023 · S-024Build the pieces. Then try a new sum.
Show why thirds and quarters need a common-sized piece. Then ask the learner to explain 1/2 + 1/3 using sixths, with the explanation kept beside the answer.
A clearer class picture.
A focused teaching response.
The proposed build covers a defined question set and marking guide, with private teacher review. It supports teaching decisions; it does not automatically label learners or alter assessment records.
Keep every answer visible.
- Correct result
- 12
- Added denominators
- 6
- Equivalent-fraction error
- 4
- Blank response
- 2
“2/7”
A likely denominator error, but the method is not written. Confirm the reasoning before accepting the group.
What changed? What stayed the same?
Explain the pieces first.
Then solve a new sum.
How we’d make it happen.
Connect each answer to its question
Read the permitted responses and marking guide. Preserve original text and distinguish an unreadable response from an actual blank.
Review the suggested patterns
Group similar reasoning and show example passages. Teachers confirm, split, or move responses before the report is used.
Choose the next teaching step
Draft a targeted explanation and follow-up task. The teacher approves its wording and decides how to use it without automatically changing grades.
Where AI helps
Document and language models read supplied responses and propose evidence-based error categories against the marking guide. Code reconciles counts and preserves the link to every original answer. Short, blank, unclear, and unreadable responses remain distinct. The tool does not infer a learner’s ability or diagnose a learning condition from a single answer.
What people approve
Teachers verify every suggested group, resolve ambiguous reasoning, decide whether a follow-up is appropriate, and retain control of grading. The institution sets access, de-identification, and retention rules for learner work. Reports describe responses to a specific task, not permanent labels for students.
What we’d need to start
- Permitted, de-identified written answers linked to the exact question and assessment version
- The marking guide, acceptable methods, and teacher-defined review categories
- Privacy and retention requirements, teacher reviewers, and the intended follow-up format
How we’d test it
Compare suggested categories with teacher-reviewed examples, including correct alternative methods, unexplained wrong answers, blanks, and unclear scans. Check false groupings and missed patterns, reconcile every response to the total, and verify that teachers can inspect the source. Evaluate a follow-up separately before claiming improved understanding.
Prepared for Education & learning. We would agree on the work after reviewing your brief, materials, and requirements.
This is a proposed project, not completed client work. The visuals and numbers show example outputs, not measured results.
