This volume contains papers from the 1st Workshop on Countering Disinformation with Artificial Intelligence (CODAI), held at the European Conference on Artificial Intelligence (ECAI) 2024.
Social media platforms which have been designed primarily to allow users to create and share content with others, have become integral parts of modern communication, enabling people to connect with friends, family as well as for broadcasting information to a wider audience. On one side these platforms provide an opportunity to facilitate discussions in an open and free environment. On the flip side, new societal issues have started emerging on these platforms. Among all the issues, the topic of misinformation has been prevalent on these platforms. The term misinformation is an umbrella term which encompasses various entities such as fake news, hoaxes, rumors to name a few. While misinformation refers to non-intentional spread of non-authentic information, the term disinformation points to spreading of a piece of inauthentic information with certain malign intentions.
Initially, researchers have mainly focused on identifying and characterizing misinformation using text based techniques through traditional and advanced NLP techniques. However, with the advancement of techniques and availability of various AI tools, the (mis)information has started appearing in the form of multimodality. For example, a piece of image with incorrect text embedded on it or a morphed video with audio. In addition, the topic of misinformation has impacted individuals and communities from various domains such as medical, political, entertainment, business, etc. This calls for combining forces from different domains. In other words, to counter misinformation computer scientists need to work with domain specialists. To understand the intention a psychologist's inputs can also be vital to understand the reasons for the spreading of misinformation. To summarize, a holistic view is needed to counter the menace of misinformation spread on online social media platforms.
The goal of this workshop is to bring together researchers interested in various domains to not only present their works but also to provide an ecosystem for discussing ideas that facilitate countering the spread of misinformation. We received a total of 17 submissions to the main workshop, of which seven were accepted as oral presentations. Finally, the workshop will feature two distinguished keynote speakers: Paolo Rosso, Universitat Politècnica de València, and David Camacho, Universidad Politécnica de Madrid, Spain.
Paolo Rosso Universitat Politècnica de València Abstract: The rise of social media has offered a fast and easy way for the propagation of disinformation and conspiracy theories. Despite the research attention that has received, disinformation detection remains an open problem and users keep sharing texts that contain false statements. In this keynote I will briefly describe how to go beyond textual information to detect disinformation, taking into account also affective and visual information because providing important insights on how disinformation spreaders aim at triggering certain emotions in the readers. I will also describe how psycholinguistic patterns and users' personality traits may play an important role in discriminating disinformation spreaders from fact checkers. Moreover, I will comment on some studies on the propagation of conspiracy theories. In the framework of the PAN Lab at CLEF, we are organising a challenge on oppositional thinking analysis to discriminate between conspiracy narratives and critical thinking. This distinction between critical and conspiracist narratives is vital because considering a message as conspiratorial when it is only oppositional to mainstream views could start a psychosocial process that drives into the arms of the conspiracy communities those who were simply critical about controversial topics such as vaccination or climate change. Most of the work was done in the framework of IBERIFIER, the Iberian media research and fact-checking hub on disinformation funded by the European Digital Media Observatory, and the research projects XAI-DisInfodemics (eXplainable AI for disinformation and conspiracy detection during infodemics), and FAKEnHATE-PdC (FAKE news and HATE speech).
Bio: Paolo Rosso is Full Professor of Computer Science at the Universitat Politècnica de València, Spain. His current research interests fall mainly in the area of detection of harmful information in social media, both fake news and hate speech. He is the principal investigator of two related projects: XAI-DisInfodemics on eXplainable AI for disinformation and conspiracy detection during infodemics (PLEC2021-007681), and FAKEnHATE-PdC on FAKE news and HATE speech (PDC2022-133118-I00), both funded by the Spanish Ministry of Science, Innovation and Universities, and by European Union NextGenerationEU/PRTR. He collaborated with the Spanish National Security Department and with the Science and Tech.
Intelligence: boundaries, threats, and trends David Camacho Universidad Politécnica de Madrid Abstract: Disinformation (and more generally misinformation) is spreading everywhere online, causing problems for individuals, societies, and countries. This unchecked dissemination of falsehoods, has nurtured an environment ripe for the proliferation of rumors, propaganda, and hoaxes, exacting a toll on the economic, political, and public health realms, among many other aspects in our daily lives. Confronting this multifaceted adversary demands a united front, drawing upon the collective wisdom and resources of diverse stakeholders including individuals, media entities, governmental bodies, technology firms, and scholars. This keynote endeavours to illuminate the intricate contours of this challenge, delving into some popular Computational techniques such as Machine Learning and Graph Computing as a new set of weapons in the battle against misinformation. Focused primarily on three domains, Natural Language Processing (NLP) and Multimodal Deep Learning (MDL) and Social Network Analysis (SNA), our discourse aims to unveil the potential of these techniques in discerning truth from falsehood. Within the realm of NLP/MDL and SNA, particular attention will be devoted to the facter-check architecture, a novel framework that through the use of ensembles and deep learning techniques based in Transformer technology, enables the identification and tracking of misleading content across the vast expanse of online social networks.