RefScore AI is a machine learning-based system designed to estimate the quality of academic references using publication metadata.
The system analyzes features such as:
- Publication year
- Citation count
- Publisher
- Number of authors
- Document type
It provides a quality score that helps students and researchers select more reliable and relevant academic sources.
👉 https://ref-rank-score.base44.app/
Selecting high-quality academic references is often time-consuming and subjective.
Many students rely on:
- citation count alone
- outdated sources
- unreliable publishers
This project aims to provide a structured and data-driven approach to evaluating reference quality.
RefScore AI uses machine learning to predict a quality score based on metadata features.
The system:
- Processes input reference data
- Extracts relevant features
- Applies a trained model
- Outputs a quality score
- Academic reference scoring system
- Metadata-based evaluation
- Simple web interface
- Real-time scoring
- Lightweight and fast
- React
- Vite
- Tailwind CSS
- Python (model training and logic)
RefScore-AI/ │ ├── src/ # Frontend source code ├── public/ # Static assets ├── index.html ├── package.json ├── tailwind.config.js ├── vite.config.js │ ├── Research Paper-3.pdf ├── Kamal-Zada_Mustafa_IT2305_research_project.pdf │ └── README.md
- User inputs reference metadata
- System processes input features
- Machine learning model evaluates the reference
- A quality score is generated
git clone https://github.com/kamalzada37/RefScore-AI.git
cd RefScore-AI
npm install
npm run dev
This project was developed as part of a research study focused on improving academic reference selection using machine learning techniques.
- Integrate real citation databases (Google Scholar, Scopus)
- Improve model accuracy with larger datasets
- Add NLP-based analysis of abstracts
- Build recommendation system for references
Academic + Applied Machine Learning Project
MIT License