CEDMO becomes part of the European Digital Media Observatory (EDMO) network of hubs on digital media that fight against disinformation across Europe. AIC researchers contribute to the EU-funded project with the use of artificial intelligence to detect disinformation and map its dissemination.
The Central European project called Central European Digital Media Observatory (CEDMO), which succeeded in the pan-European competition, will be based in Prague and its leaders will be experts from the Czech Technical University (AI Center FEE CTU) and the Charles University. The goal of CEDMO is interdisciplinary research into disinformation, including its detection using the elements of artificial intelligence, strengthening digital and media literacy and the development of fact-checking platforms. The project will become part of the European Digital Media Observatory network and will be supported by the National Recovery Plan of the Czech Republic. It is also the first step in developing a European Centre of Excellence in AI.
AI-based fact-checking support
Unique to the CEDMO project is the use of artificial intelligence to map the dissemination and detect disinformation. The development and research of technological and AI-based fact-checking support is our primarily objective in this project. Our researchers led by Luboš Král participate on this task together with the Slovakian KInIT (Kempelen Institute of Intelligent Technologies).
The research on our part will tap on the potential of human and AI cooperation, so called human in the loop. It will include AI-based methods like: 1) spotting “fact-check-worthy” articles/claims, 2) detection of potential fact-check duplicates or 3) recommendation of reliable information sources to help the fact-checker. These methods will enable automated screening of monitored data sources by filtering or re-ordering the incoming content items. They will also help additional information that will make the human fact-checking faster, for example by highlighting likely false claims, listing similar proven toxic sources and content, listing relevant trusted information sources etc.
The second pillar of our engagement will go beyond manual human work in an attempt to automatize the fact-checking process itself. Automation concerns in particular methods of mass data processing using AI methods, mainly natural language processing and natural language understanding, including also methods like word2vec. We will use them to provide content credibility detection, facts retrieval, information bias detection, perplexity and news informational distortion development over the time, allowing also their anomaly and manipulation detection. The outcomes of this activity will be a functional prototype of platform, supporting tools and AI methods and technical and research reports on tools’ and AI methods’ design and evaluation.
During the first two years of its existence, the CEDMO project will be supported from the CEF Telecom 2020-EU-IA-0267 call and also from the funds of the National Recovery Plan of the Czech Republic. This will cover the participation of Czech members of the consortium (i.e. not only CTU and Charles University, but also the fact-checking platform Demagog.cz) and especially the continuation of the project after the year 2023 and the creation of a unique CEDMO Index, which will allow for long-term research into the impact of disinformation on the Czech population and the media.