CERTH (the vera.ai project coordinator) is one of the largest Greek research centres and among the Top-10 EU's Research Centres in attracting research grants. CERTH is represented by the Information Technologies Institute's Media Verification team which brings experience in AI for multimedia and social network analysis from numerous projects on social media verification. CERTH will research image and video analysis; visual manipulation detection, especially deepfake detection and image forensics; reverse image and video search; methods for extracting contextual knowledge from visual content; and social network analysis and visualisation methods.
The Natural Language Processing (NLP) group at the University of Sheffield (USFD) is one of the largest and most successful research groups in text and social media analysis and language processing in Europe. USFD led the PHEME project, which was the first to study AI-based methods for rumour analysis. Apart from the Coordinator role, USFD will provide overall scientific direction and work on the definition, development and implementation of open-source language-based methods and tools for detecting disinformation.
The University of Urbino Carlo Bo (UNIURB) is an Italian university located in Urbino, founded as an academic institution in 1506. UNIURB is traditionally renowned for teaching and research in humanities, but recently also for life, environmental and earth sciences. The team of scholars involved in vera.ai has an outstanding track record of publications and research in the field of network science, media and political communication. In the project UNIURB will be focusing on social media analysis and online behavioural modelling.
The Institute for Digital Media Technology (IDMT) of Fraunhofer brings expertise from many relevant research fields into the project, such as audio-visual signal analysis and machine learning, media security and privacy, acoustics and 3D audio, audio forensics, cross-modal analysis and recommendation, and more. IDMT will mainly contribute to tasks on explainable audio analysis and provide tools for audio manipulation and deepfake audio detection.
The University of Amsterdam (UvA) is one of the largest research universities in Europe. The Media Studies department has been ranked 1st by QS World University Rankings from 2018 to 2021. Its Digital Methods Initiative (DMI) group has pioneered issue mapping with the Issue Crawler. DMI has built large social media data collections, developed social science and interpretative techniques which underpin their cross-platform analyses of content moderation, misinformation, extreme speech, toxicity, radical subcultures, etc. UvA will utilise the novel vera.ai tools to deepen its studies of online disinformation and its social science researchers will lead the co-creation activities.
The Kempelen Institute of Intelligent Technologies (KInIT) is an independent research organisation, working on AI, data science, machine learning, natural language processing, information security, and software engineering. Misinformation has been a core area of research in the past years, as well as AI ethics including practical experience with ethical assessment of AI products. KInIT will contribute research capabilities in machine learning, analysis of web and user behaviour, natural language processing, and algorithm auditing to the project. KInIT will specifically work on multilingual approaches, with particular focus on Western Slavic languages through cross-lingual transfer learning.
The University of Naples Federico II (UNINA) is one of the largest research institutions in Italy and the largest university of southern Italy. The Image Processing Research Group (GRIP) brings over 20 years of research in processing images and video, including image and video forensics, biometric liveness detection, image segmentation, classification and restoration into vera.ai. Focusing on multimedia forensics, GRIP will provide their expertise in Multimedia Forensics for video and image analysis using deep learning approaches for image and video manipulation detection and localisation, audio-visual analysis for video deepfake detection, and near-duplicate image and video detection.
The ENS Centre Borelli is part of the department of mathematics of Ecole Normale Supérieure Paris-Saclay (ranked 1st for mathematics in the Shanghai classification). It is internationally recognised for several inventions in mathematical image processing and selected by the French army research agency to devise new classes of image forgery detectors. ENS will provide their expertise in signal processing methods for image and video analysis, detecting the traces and parameters of image and video tampering. ENS will furthermore apply its probabilistic detection theory to associate with each image or video anomaly a probability of false alarms.
Athens Technology Centre (ATC) is an IT SME offering solutions and services for the media market, with particular focus on tools for analysing misinformation. ATC provides the technical infrastructure that supports EDMO's communities of fact-checkers and researchers in studying and analysing disinformation. ATC has products that are used by broadcasters, news agencies, and journalists worldwide to verify and debunk online content. In vera.ai, ATC will lead the integration activities, working on integrating the AI components into the Truly Media verification platform used by EDMO. ATC will also lead the Exploitation activities and play a major role in clustering activities.
Sirma AI (trading as Ontotext - ONTO) offers world-leading semantic technology with a unique combination of text analytics, content enrichment, knowledge discovery and semantic search, and an enterprise-grade graph database engine (GraphDB™). Ontotext will leverage its AI and knowledge graph expertise to advance its Database of Known Fakes towards more EU countries and languages. ONTO will also implement deep-learning based multilingual semantic search, create semantic models of disinformation campaigns and narratives, and provide insights about major actors, spread patterns and impact of disinformation.
Agence France Press (AFP) is a global news agency, working in six languages, spanning video, text, photos, graphics and online services. Since 2017 AFP grew to become the leading global fact-checking organisation, operating a global network of nearly 90 in-house journalists and editors focused on digital investigation and false information debunking. AFP is also the lead developer and designer of the widely used, award-winning InVID-WeVerify plugin. AFP will drive user needs, requirements and evaluation in vera.ai, with a strong focus on the latter. Integration of AI tools into the verification plugin will be the other important focus, as well as playing a leading role in innovation management, clustering, and training activities.
Deutsche Welle (DW) is Germany's international public service broadcaster, producing and distributing content in 32 languages. The team in charge of all vera.ai activities is based in the Research & Cooperations Projects Unit. DW's main focus in vera.ai will be on use cases, requirements and evaluation; and on leading the Dissemination task, bringing in its expertise in communication and outreach. DW will also bring first hand experiences of media professionals, a core target of this call, into the project. All this is to ensure that work and activities are demand-driven, meet clear professional user needs, have highest usage potential and are communicated professionally and appropriately to the respective target audiences.
EU DisinfoLab (EUDL) is an independent NGO and SME focused on investigating disinformation campaigns targeting the EU, its member states, institutions, and core values. EUDL maintains regular contacts with policy makers and has established collaboration with international media and international organisations (WHO, UNESCO, NATO, etc.). EUDL will lead the task of establishing and maintaining links and relationships with other organisations and initiatives and will contribute through its wide community and experience in detecting and analysing disinformation campaigns.
The European Broadcasting Union (EBU) represents 115 public media organizations in 56 countries and 34 Associates. EBU Members operate nearly 2,000 television and radio channels, which together reach audiences of more than one billion people, broadcasting in 160 languages. The EBU delivers a variety of services to its members including the fact-checking and verification service Eurovision Social Newswire. It also runs a number of transversal strategic initiatives such as the interdisciplinary AI and Data Initiative (AIDI) on the challenges of AI and data, develops and incubates media technology, and provides training for media professionals at its EBU Academy. Its involvement in vera.ai will ensure widest possible uptake and involvement of the European (public service) media sector, making significant contributions to requirements analysis and dissemination, communication and clustering.