The MUSMA project will make a radical change by investigating and developing innovative analysis models that can jointly process and understand textual and visual content simultaneously. MUSMA will enable:

• processing, understanding, selection and monitoring of online content, to extract relevant information on a set of specific topics or subjects;

• misinformation and manipulated content detection, through the development of AI techniques specifically designed to identify the network-wide phenomena (e.g. emerging communities and viral content) and credibility of online content;

• analysis of the main drivers of information consumption, the dynamics of information and misinformation flow and ranking information sources according to their topical influence.

At the core of the project lies a new unifying synergy between Network Science, NLP and CV, using supervised neural networks (going beyond convolutive autoencoders, Transformer-based NNs, Capsules and graph-based networks) and symbolic representations.