The FERMI project is an ambitious undertaking with a clear-cut objective: to create and validate technologies that address the ever-evolving challenge of disinformation and its tangible consequences. The project is organised around two key technical work packages: the FERMI Core Technology Offerings and the User Interfaces and Platform Integration. These facets are intrinsically interconnected in their pursuit of delivering the Minimum Viable Product (MVP) of the FERMI Platform within the first year of the project.
Progress so far
Over the past year, significant strides have been made in the development of critical FERMI core technologies. These groundbreaking innovations includes (i) the Dynamic Flows Modeller (DFM) – Crimes Impact Predictor an AI-powered tool that links online disinformation and fake news (D&FN) to offline crimes, enhancing our understanding of this relationship and helping uncover societal and cultural factors behind disinformation as it enables early identification of disinformation-induced offline risks, empowering law enforcement agencies (LEAs) to implement timely security measures; (ii) the Disinformation Analyser that analyses D&FN campaigns mapping them to main actors/accounts, classifying them as persons or bots and providing an influence index; (iii) the built on ISO standards and scientific literature Community Resilience and Risk Management model that assesses a community's ability to rebound after D&FN exposure using Multi-Criteria-Decision-Making to consider behavioural, socioeconomic, and threat factors; (iv) the Behaviour Analyser that generates indicators exerting an influence on the probability of extremist criminal activities stemming from online engagements that contribute to this phenomenon encompass, among other variables, the level of media literacy and educational attainment; (v) the Socioeconomic Analyser that quantifies crime related costs and can even differentiate them based on motivation, whether it be right-wing, left-wing, religious, or foreign; (vi) the Swarm Learning Module which allows the training of AI models tailored to each LEA’s data to be used by the Dynamic Flows Modeller (DFM) – Crimes Impact Predictor and (vii) the Sentiment Analysis Module which utilises the BERT language model and delves into the emotional polarity of D&FN related social media posts, specifically on platforms such as X as its primary function lies in corroborating the origins of counterfeit news by scrutinising the sentiments articulated within tweets.
Collaboration is Key
The FERMI project places significant emphasis on collaborative efforts among partners to maximise the effectiveness of the integrated technologies and enhance result accuracy. Furthermore, it incorporates a state-of-the-art Continuous Integration/Continuous Development (CI/CD) pipeline, ensuring both the quality and automated delivery of updated components. Simultaneously, rigorous scrutiny has been applied to the design process of the FERMI front-end to ensure it aligns seamlessly with overarching project objectives. This design approach is rooted in user-centric principles and integrates FERMI's advanced Artificial Intelligence (AI) capabilities that not only prioritises transparency but also ensures comprehensibility in decision support, providing a holistic and user-friendly experience.
Concurrently, groundwork is being laid for the Socioeconomic Disinformation Watch, building upon the foundation established by the development of FERMI's core technologies. This module aims to play a crucial role in addressing D&FN challenges within the European Union (EU) as it is designed to empower LEAs and security professionals to effectively combat evolving threats posed by disinformation campaigns. Through systematic monitoring of trending topics, the Socioeconomic Disinformation Watch aims to serve as an early warning system, detecting potential D&FN risks in areas of public interest. This proactive approach equips LEAs and national ministries to respond promptly to emerging threats, thereby enhancing the security of EU society.
Moving forward, the project maintains its primary emphasis on continuous improvement, with a specific emphasis on achieving and validating two integrated iterations of the FERMI platform. The first integrated version is scheduled for delivery at month 18, while the final version is targeted for completion at month 30 of the project timeline. These endeavours serve as a testament to the resolute commitment of project partners in the pursuit of the overarching mission and objectives of the FERMI project.