Starke and Zlabginer’s Imagetwin: Protecting the integrity of research with AI
Winner Interview 2023: Science & Innovation Management
Introducing Imagetwin, the revolutionary solution by Patrick Starke and Markus Zlabinger that ensures the integrity of research by utilizing image detection technology to prevent academic mistrust and fraudulent publications. Rampant image-related problems like manipulation and plagiarism compromise the authenticity of scholarly work. Imagetwin breaks this wall, offering an automated image detection technology that meticulously identifies incorrect images, enhancing the trustworthiness of research. After six years of dedicated development, this AI-driven tool provides a comprehensive solution, bolstering the credibility of scientific publications and preventing the dissemination of flawed research. Beyond academia, this innovation safeguards resources, reputation, and progress, fostering a culture of trust and integrity in the scientific community.
Which wall does your research break?
Imagetwin breaks the wall of research misconduct and ensures the integrity of science by offering a solution to detect fraudulent images in academic articles. A significant proportion of academic manuscripts contain image-related problems like manipulation and plagiarism. Unfortunately, even though automated solutions for text plagiarism have existed for many years, there was no such solution for images in the past and the checking was done manually or not at all. But manually checking images for integrity issues is time-consuming, and expensive, and besides the lack of trained experts, many problems remain undetected.
Due to the overall improvement of AI over the previous years, it is now possible to offer the first automated solution for the detection of incorrect images in scientific papers. The primary challenge during the development of Imagetwin was creating a robust and accurate algorithm that is not only able to detect duplicated and manipulated images but also to be fast and efficient without producing a high number of false positive results. To fine-tune the technology for this matter was a huge challenge. This endeavor demanded over six years of dedicated research and development.
The overarching vision of the Imagetwin project is to contribute to a world free from fraudulent practices, particularly targeting paper mills in the academic and scientific communities. Paper mills are entities that produce and sell pre-written academic papers, facilitating academic dishonesty. By automating the detection and verification of image integrity, we aim to prevent dishonest practices where researchers or publishers use manipulated or plagiarised images to deceive the public.
The benefits of the project extend far beyond the academic sphere. Every year, substantial resources are wasted due to flawed research and publications, and problematic images play a central role in this issue. By providing an automated and efficient solution for detecting image integrity issues, Imagetwin can significantly reduce the occurrence of fraudulent publications and research. This will enhance the overall quality and reliability of scientific publications, saving time, money, and effort for researchers, institutions, and the public.
What inspired or motivated you to work on your current research or project?
It all began as a student project at the Technical University of Vienna in 2016, and by chance, we stumbled upon the issue of image-related problems in research. As we delved deeper into the subject matter, we realised the magnitude of the problem. Due to the fascination, Markus dedicated his master's thesis to this topic. The more he learned, the clearer it became that an AI-based software solution could be a great asset in combating the problem of research misconduct.
To gain valuable insights and guidance, we engaged in discussions with leading experts in the field, such as Elisabeth Bik and Jana Christopher. These interactions confirmed that there was indeed a pressing demand for automated software that could address image-related problems, similar to what already exists for text-related issues.
However, the ultimate driving force behind our work on Imagetwin goes beyond the technical aspects and industry demands. Our main motivation lies in our belief that a world with high-quality research is essential for solving society's greatest challenges, such as climate change and global health. We recognise that reliable and trustworthy scientific knowledge is the key to making informed decisions and driving positive change on a global scale.
Our vision for Imagetwin is not just about improving academic and scientific practices; it's about creating a positive impact on society as a whole. We strive to contribute to a world where research integrity is upheld, where fraudulent practices are minimised, and where scientific advancements are fueled by accurate, credible, and reliable information.
In what ways does society benefit from your research?
Society benefits from our research project in several ways. With the ever-increasing number of academic publications, ensuring good quality standards has become a challenging task. Imagetwin addresses this issue and actively supports universities, publishers, and scientists worldwide in their mission to uphold the quality and trustworthiness of scientific research.
One of the most significant contributions of our research is the ability to identify image-related problems like plagiarism and manipulation. By doing so, we can substantially reduce the prevalence of unethical and flawed papers. This, in turn, leads to a higher overall quality of research being disseminated.
The early detection of fraudulent academic papers before publication is a crucial task. By preventing flawed research from being published, our software helps to mitigate the occurrence of misdirected follow-up research. This not only saves researchers from pursuing unfruitful paths but also allows resources to be directed toward more promising and impactful areas of study.
Furthermore, Imagetwin plays a vital role in safeguarding the reputation of universities, publishers, and scientists. By proactively screening out problematic papers before they are published, it helps prevent the need for retractions months or years down the line. Retractions often result in significant reputational damage and incur substantial costs for the necessary investigations. Preventing such retractions saves resources and preserves the credibility of the academic community.
Ultimately, the societal benefit of our research lies in fostering a culture of trust and integrity in scientific research. Identifying fraudulent research contributes to a world with higher research integrity. This, in turn, enhances the reliability of scientific knowledge, making it a key foundation for academic progress.
