AI-Based Question Paper Generator with CO–Bloom Mapping
Inspiration
Preparing university question papers is a time-consuming process for faculty members. Creating balanced assessments that align with Course Outcomes (COs) and Bloom's Taxonomy, while avoiding repeated questions and ensuring proper difficulty distribution, is challenging. We wanted to simplify this process by building an AI-powered platform that automates question paper generation while maintaining academic quality and outcome-based education standards.
What it Does
Our system helps faculty members create high-quality question papers by:
- Managing a centralized question bank.
- Mapping each question to the appropriate Course Outcome (CO) and Bloom's Taxonomy level.
- Automatically generating balanced question papers based on predefined rules.
- Detecting repeated or similar questions from previous examinations.
- Generating a draft scheme of evaluation for easier assessment.
- Providing AI/ML-based appraisal scoring and recommendations to improve question quality and assessment coverage.
How We Built It
We designed the system using a modular architecture consisting of:
- Frontend: React.js for an intuitive user interface.
- Backend: Java (Spring Boot) to manage business logic and APIs.
- Database: MongoDB for storing question banks, CO mappings, Bloom levels, and generated papers.
- AI/ML: Machine learning techniques to evaluate question quality, detect repetition, and recommend improvements.
- Version Control: Git and GitHub for collaborative development.
The workflow begins with faculty uploading or creating questions. Each question is tagged with its CO and Bloom level, after which the AI engine generates a balanced paper while checking constraints such as repetition, coverage, and difficulty.
Challenges We Ran Into
Some of the biggest challenges included:
- Designing an efficient question-tagging system.
- Balancing question distribution across COs and Bloom levels.
- Detecting duplicate and semantically similar questions.
- Generating papers that satisfy multiple academic constraints simultaneously.
- Developing meaningful AI recommendations for improving assessment quality.
What We Learned
This project strengthened our understanding of:
- Outcome-Based Education (OBE)
- Bloom's Taxonomy
- AI-assisted educational technologies
- Machine Learning for text analysis
- Full-stack development with React, Java, and MongoDB
- Database design and API integration
What's Next
We plan to enhance the platform by adding:
- NLP-based semantic similarity detection.
- Automatic question generation using Large Language Models (LLMs).
- Support for multiple university regulations and examination formats.
- Faculty analytics dashboards.
- Difficulty prediction and adaptive assessment generation.
- Integration with Learning Management Systems (LMS).
Our goal is to make assessment creation faster, smarter, and more reliable, allowing educators to focus more on teaching while ensuring fair, balanced, and outcome-driven evaluations.
Log in or sign up for Devpost to join the conversation.