A wealthy friend of mine asks for a vacation credit card to his bank, to discover that the credit he is offered is very low. The bank teller cannot explain why. My stubborn friend continues his quest for explanation up to the bank executives, to discover that an algorithm lowered his credit score. Why? After a long investigation, it turns out that the reason is: bad credit by the former owner of my friend’s house. Black box AI systems for automated decision making, often based on ML over (big) data, map a user’s features into a class or a score without explaining why. This is problematic for lack of transparency, but also for possible biases inherited by the algorithms from human prejudices and collection artefacts hidden in the training data, which may lead to unfair or wrong decisions. I strive for solutions of the urgent challenge of how to construct meaningful explanations of opaque AI/ML systems, introducing the local-to-global framework for black box explanation, articulated along 3 lines: a) the language for explanations in terms of expressive logic rules, with statistical and causal interpretation; b) the inference of local explanations for revealing the decision rationale for a specific case; c), the bottom-up generalization of many local explanations into simple global ones. An intertwined line of research will investigate both causal explanations, i.e., models that capture the causal relationships among the features and the decision, and mechanistic/physical models of complex system physics, that capture the data generation mechanism behind specific deep learning models. I will also develop: an infrastructure for benchmarking, for the users' assessment of the explanations and the crowdsensing of observational decision data; an ethical-legal framework, for compliance and impact of our results on legal standards and on the “right of explanation” provisions of the GDPR; case studies in explanation-by-design, with a priority in health and fraud detection.

SoBigData++ strives to deliver a distributed, Pan-European, multi-disciplinary research infrastructure for big social data analytics, coupled with the consolidation of a cross-disciplinary European research community, aimed at using social mining and big data to understand the complexity of our contemporary, globally-interconnected society. SoBigData++ is set to advance on such ambitious tasks thanks to SoBigData, the predecessor project that started this construction in 2015. Becoming an advanced community, SoBigData++ will strengthen its tools and services to empower researchers and innovators through a platform for the design and execution of large-scale social mining experiments. It will be open to users with diverse background, accessible on project cloud (aligned with EOSC) and also exploiting supercomputing facilities. Pushing the FAIR principles further, SoBigData++ will render social mining experiments more easily designed, adjusted and repeatable by domain experts that are not data scientists. SoBigData++ will move forward from a starting community of pioneers to a wide and diverse scientific movement, capable of empowering the next generation of responsible social data scientists, engaged in the grand societal challenges laid out in its exploratories: Societal Debates and Online Misinformation, Sustainable Cities for Citizens, Demography, Economics & Finance 2.0, Migration Studies, Sport Data Science, Social Impact of Artificial Intelligence and Explainable Machine Learning. SoBigData++ will advance from the awareness of ethical and legal challenges to concrete tools that operationalise ethics with value-sensitive design, incorporating values and norms for privacy protection, fairness, transparency and pluralism. SoBigData++ will deliver an accelerator of data-driven innovation that facilitates the collaboration with industry to develop joint pilot projects, and will consolidate an RI ready for the ESFRI Roadmap and sustained by a SoBigData Association.

Recently, there has been a paradigmatic shift in experimental molecular spectroscopy, with new methods focusing on the study of molecules embedded within complex supramolecular/nanostructured aggregates. In the past, molecular spectroscopy has benefitted from the synergistic developments of accurate and cost-effective computational protocols for the simulation of a wide variety of spectroscopies. These methods, however, have been limited to isolated molecules or systems in solution, therefore are inadequate to describe the spectroscopy of complex nanostructured systems. The aim of GEMS is to bridge this gap, and to provide a coherent theoretical description and cost-effective computational tools for the simulation of spectra of molecules interacting with metal nano-particles, metal nanoaggregates and graphene sheets. To this end, I will develop a novel frequency-dependent multilayer Quantum Mechanical (QM)/Molecular Mechanics (MM) embedding approach, general enough to be extendable to spectroscopic signals by using the machinery of quantum chemistry and able to treat any kind of plasmonic external environment by resorting to the same theoretical framework, but introducing its specificities through an accurate modelling and parametrization of the classical portion. The model will be interfaced with widely used computational chemistry software packages, so to maximize its use by the scientific community, and especially by non-specialists. As pilot applications, GEMS will study the Surface-Enhanced Raman (SERS) spectra of systems that have found applications in the biosensor field, SERS of organic molecules in subnanometre junctions, enhanced infrared (IR) spectra of oligopeptides adsorbed on graphene, Graphene Enhanced Raman Scattering (GERS) of organic dyes, and the transmission of stereochemical response from a chiral analyte to an achiral molecule in the vicinity of a plasmon resonance of an achiral metallic nanostructure, as measured by Raman Optical Activity-ROA

In Humanities, written correspondence is a fundamental object of study. The current IT revolution provides means for reassembling and disseminating this precious literary heritage for the first time, while at the same time fostering new forms of scholarly cooperation. As letters create links between people, the value of documents and data increases when different corpora of written correspondences are interconnected and linked. Despite the strong need for common standards and common tools to exploit these connections and to reconstruct the discourse of the Republic of Letters, there is still no simple framework in the current panorama for publishing digitized correspondence corpora. The OPenPal project, led by Prof. Ghelardi, aims to develop a prototype system to manage, publish, export and visualise corpora of various types of correspondence in innovative ways: the idea is therefore to move from the EU-funded project EUROCORR (storing and displaying the critical edition of the correspondence to Jacob Burckhardt (1842-1897) – http://fe.burckhardtsource.org/ –) and to generalise both workflow and data model to new correspondence corpora, by refactoring and redesigning the architecture of Burckhardtsource. The expected outcomes of the OPenPal project will be to produce a prototype to publish digital correspondences (i.e. manuscripts, texts and data) and a toolkit to manage these special editions. OPenPal prototype will enable users to browse texts and data through guided paths, infographic visualisations, and will make it possible to browse a unique space of linked knowledge and trigger social mechanisms for discussion and debate on letters, thus opening from a niche of scholarly content to a wider audience. As a whole, this process will identify a sustainable business model for the new tool, based on linked data technologies in correspondence digital editions.