Scientific writing often contains nuanced expressions of uncertainty—phrases that convey doubt, probability, or conditionality. Identifying and understanding these uncertainties is critical for researchers, reviewers, and readers to accurately interpret scientific claims. The UnScientify project addresses this challenge by leveraging advanced natural language processing to detect and classify scientific uncertainty in texts.

About the Demo Application

Funded by French ANR JCJC 2021 - 2025, ANR-21-CE38-0003-01, as a part of InSciM Project, the UnScientify demo application offers an interactive interface for users to explore the system’s capabilities firsthand. Designed with simplicity and user experience in mind, the application allows users to input any scientific text and receive immediate annotation of uncertainty expressions present in the content. This real-time feedback provides a practical way to understand how the model interprets language nuances.

The interface features a clean text box for input and dynamically displays annotated results, highlighting uncertain statements directly within the text. This interactive experience supports users from varied backgrounds, from domain experts to curious learners, by making complex model outputs accessible and interpretable.

Access the demo application here: UnScientify Demo

Features of the UnScientify Demo

• Real-time annotation: Instantly identifies uncertain phrases and marks them within the input text.

• User-friendly interface: Minimalist and intuitive design suitable for quick testing and exploration.

• Accessibility: Hosted on GitHub and Streamlit for straightforward usage without complicated setup.

Background and Research

UnScientify builds on a foundation of extensive research and development in scientific uncertainty detection. Key publications detailing the methodology, the system design and its diverse implementations include:

• Panggih Kusuma Ningrum, Phillip Mayr, Nina Smirnova, Iana Atanassova (2025). Annotating Scientific Uncertainty: A comprehensive model using linguistic patterns and comparison with existing approaches. In : Journal of Informetrics (2025).

• Panggih Kusuma Ningrum, Nicolas Gutehrlé, Iana Atanassova (2025). Etudier l’incertitude dans les articles scientifiques : mise en perspective d’une méthode linguistique. In : Extraction et Gestion des Connaissances 2025, Thomas Guyet, Baptiste Lafabrègue, Aurélie Leborgne, Jan 2025, Strasbourg, France.

• Iana Atanassova, Panggih Kusuma Ningrum, Nicolas Gutehrlé, Francis Lareau, Christophe Malaterre (2025). Is There Life on Mars? Studying the Context of Uncertainty in Astrobiology. In: 20th International Conference of the International Society for Scientometrics and Informetrics (ISSI 2025), Yerevan, Armenia.

• Panggih Kusuma Ningrum, Iana Atanassova (2024). Annotation of scientific uncertainty using linguistic patterns. Scientometrics (2024).

• Christophe Malaterre, Iana Atanassova, Francis Lareau, and Panggih Kusuma Ningrum. Navigating the unknown: A textual analysis of uncertainty in astrobiology research. In Congress of the Canadian Society for the History and Philosophy of Science (CSHPS), 2024.

These studies highlight the broad potential for UnScientify to enhance understanding of scientific discourse across fields.

Ongoing Development

The UnScientify demo remains an evolving project with continuous improvements planned to expand its features and usability. Feedback from the user community is invaluable for refining the system and guiding future enhancements.