███████╗  ██╗       ██╗  ███████╗
██╔════╝  ██║       ██║  ██╔════╝
█████╗    ██║       ██║  ███████╗
██╔══╝    ██║       ██║  ╚════██║
███████╗  ███████╗  ██║  ███████║
╚══════╝  ╚══════╝  ╚═╝  ╚══════╝
Scientific Integrity System

ELIS - Scientific Integrity System

ELIS is a FOREVER FREE AND OPEN-SOURCE system designed to analyze the integrity of scientific data.

Our goal is to democratize access to advanced forensic tools, empowering researchers and integrity officers with robust and transparent tools to ensure the integrity of scientific records.

Currently, the system is focused on image forensics, but future versions will extend to text and statistical data analysis.

ELIS.mp4

---

Getting Started

To get ELIS running on your machine, you will need Docker Compose and Node.js.

1. Clone the repository and submodules

git clone --recurse-submodules git@github.com:researchintegrity/elis.git
cd elis
git submodule update --init --remote # ensure latest submodule versions

1.1 Fix .env

cp .env.example .env
# Edit .env to set the HOST_WORKSPACE_PATH
# >> HOST_WORKSPACE_PATH=<path/to-current-dir>/elis/system_modules/elis-frontend/workspace

2. Build the tools

This step could take some while as it will download and compile multiple models from different servers

docker compose --profile tools build

3. Launch the backend

docker compose up -d

(Production Alternative) 3. Launch the backend with multiple workers (n=5)

docker compose -f docker-compose-prod.yml up -d --scale workers=5 

4. Launch the frontend

cd system_modules/elis-frontend
npm install
npm run dev

After instalation, visit http://localhost:5173 to see the system in action!

Tip

Need more details? Check our Technical Overview for a deep dive into the architecture, manual installation, and API documentation.

---

Implemented Modules

ELIS integrates multiple specialized modules to detect manipulation.

Module	Description	Status
PDF Image Extraction	Extracts images from scientific PDF documents for analysis.	✅
Panel Extractor	Uses YOLO models to parse multi-panel figures into individual images.	✅
Watermark Removal	Removes "RETRACTED" watermarks from PDF academic articles.	✅
CBIR Search	Finds similar images across datasets.	✅
TruFor	Detects cheapfakes and image manipulations.	✅
Copy-Move Detection	Identifies duplicated regions within and across images.	✅
Provenance Analysis	Tracks reused and manipulated data across articles and datasets.	✅

TODOs

• Documentation
• UI/UX Improvement (check the front-end issues)
• AI-image detection module
• AI-text detection module
• OpenAlex API linked to ELIS

Found a bug or want a new feature, please contribute by opening a new issue!

---

Acknowledgements

The name of this system is a tribute to Dr. Elisabeth Bik, a key personality in the field of scientific integrity. Her dedication to uncovering data manipulation has inspired our work and the work of many other researchers worldwide. Learn more about her work at her blog: Science Integrity Digest

Special Thanks:

• Forensically: The Image Analysis module is deeply inspired by this project. Shoutout to Jonas Wagner.
• UNINA Image Processing Research Group: For their work on Dense-Field copy-move detection and TruFor methods.

---

License

ELIS is open-source software licensed under the AGPLv3 License.

Note: Each module integrated into ELIS has its own licensing terms. Some components may have restrictions on commercial use. Please check the LICENSE file in each individual module for specific details.

---

"No one can predict the positive shift caused by even an ant's step on the path of science and ethics."

_{Built with ❤️ for Science}