PAN at CLEF 2023

Recent technological developments and social events have contributed to a constant increase in online disinformation of various forms making it a long-lasting challenge of immense scale and complexity. Disinformation is a significant technological and social challenge, which affects key societal values including Democracy, Public Health and Peace. Apart from technology, It depends on various dimensions such as policies, business models, education, digital literacy and human behaviour. Focusing on visual disinformation, including manipulated photos/video, deepfakes, visuals out of context and false connections, a variety of approaches and tools are needed in order to address this challenge. In this talk, after a general introduction and overview, I will be presenting our lab’s efforts in this area, across three main directions: approaches which take into account content, context and network-based information. It will include media forensics, deepfake detection and reverse image and video search approaches together with tools already used by journalists and fact-checkers. Key challenges and additional aspects such as actual operational settings, human behaviour, education and policy issues will also be covered.

Detecting and Disrupting Disinformation: Social Network Analysis and Natural Language Processing In recent years, social network analysis has become an increasingly important tool in understanding and combating the spread of disinformation. By examining the patterns of information flow, we can identify the sources and pathways through which false information spreads and take steps to prevent its dissemination. However, the problem of disinformation is not limited to the propagation of already-existing false information; AI is now being used to generate convincing fake news, amplifying the spread of disinformation and making it increasingly difficult to distinguish fact from fiction. Despite the challenges posed by AI-generated false content, AI can also play an important role in the fight against disinformation. Through the use of natural language processing techniques, we can reveal key characteristics of false information, enabling us to more effectively identify and combat disinformation campaigns. Additionally, machine learning algorithms can be used to identify patterns of disinformation and to distinguish between false and genuine content. In this talk, we will explore the current state of the art in social network analysis and AI with respect to disinformation. We will present examples of different models and architectures that have been developed to combat disinformation, including case studies of real-world disinformation campaigns. By combining the strengths of social network analysis and AI (i.e., large language models), we can develop more effective tools for combating the spread of disinformation, protecting the integrity of public discourse, and upholding the principles of truth and accuracy in our information ecosystem.