Scholar Compare: AI-Powered Short Answer Grading

Inspiration

As fulltime students, we witnessed educators drowning in stacks of short answer assessments. Current educational platforms only provide auto grading for multiple choice and numeric questions, but not written responses. We are here to change that.

What it does

Scholar Compare instantly evaluates short answer questions. Teachers:

1. Set the correct answer (1-2 sentences)
2. Upload student responses via CSV
3. Receive instant accuracy judgments
4. See class-wide performance analytics
5. Export the results in a CSV format

How we built it

Frontend:

• React + TypeScript for responsive UI
• Tailwind CSS for clean, accessible design

Backend:

• Python Flask REST API
• Google Gemini API for semantic analysis
• MongoDB for evaluation history storage

Challenges we ran into

Our main challenge was with the backend in regards to learning to use a new data base (Mongo DB).

Accomplishments we're proud of

1. Processing answers quickly
2. Intuitive and easy to use interface

What we learned

1. Semantic similarity ≠ educational correctness
2. Contextual keywords matter more than exact phrasing
3. Simple tools have biggest classroom impact

What's next for Scholar Compare

Immediate:

• LMS integrations (D2L Brightspace, Google Classroom)
• A working login page
• Mobile teacher app for on-the-go grading

Future:

• Plagiarism detection
• IEP accommodation support
• District-wide analytics dashboard

"We're not replacing teachers - we're returning their weekends."

Built With

• flask
• gemini
• mongodb
• python
• react
• typescript