Skip to content

Latest commit

 

History

History
71 lines (47 loc) · 2.49 KB

README.md

File metadata and controls

71 lines (47 loc) · 2.49 KB

📚 ReadyCite

AI-Driven Fact Checking for Scientific Citations

🎯 Project Overview

ReadyCite is an innovative machine learning system designed to address a critical challenge in scientific literature: quotation errors. By automatically detecting inaccurate citations, this project aims to enhance the integrity and reliability of academic publications.

✨ Key Features

  • Automated Citation Checking: Identify citations that do not accurately support the claims made in scientific manuscripts
  • Multi-Dataset Training: Leverages diverse datasets including SciFact, Citation-Integrity, and Fever for comprehensive learning
  • Advanced NLP Techniques: Utilizes state-of-the-art natural language processing models like Longformer and BERT
  • Scalable Architecture: Designed to process massive scientific literature datasets

🔍 Problem Statement

Quotation errors are unfortunately common in scientific literature, with studies suggesting 10-20% of citations may be inaccurate. These errors can:

  • Undermine research credibility
  • Propagate misinformation
  • Waste researchers' time in tracing citation origins

🛠 Technical Approach

The project is structured in three progressive phases:

  1. Initial Model Training

    • Use preprocessed datasets with abstracts and evidence sentences
    • Develop baseline citation accuracy classification

  2. Evidence Retrieval Pipeline

    • Expand from abstracts to full-text PDF processing
    • Implement advanced sentence retrieval techniques (BM25, MonoT5 reranking)

  3. Large-Scale Deployment

    • Cloud deployment
    • Processing extensive databases like bioRxiv

📊 Data Sources

  • SciFact: 1,409 scientific claims verified against 5,183 abstracts
  • Citation-Integrity: 3,063 citation instances from biomedical publications
  • Fever: 185,445 claims from Wikipedia (pre-training)

🚀 Planned Deliverables

  • Web application for citation accuracy verification
  • Open-source machine learning models
  • Detailed research documentation

🧰 Technologies

  • Python
  • PyTorch
  • Transformers (BERT, Longformer)
  • Natural Language Processing Libraries
  • Machine Learning Frameworks

📋 Project Status

Current Phase: Model Training and Validation

  • Reproducing Citation-Integrity model
  • Experimenting with training epochs and datasets

🤝 Contributing

Interested in improving scientific literature integrity? Contributions are welcome! Please read our contribution guidelines.

📞 Contact

Jeffrey Dick - Project Lead