Maximising opportunities and minimising risks associated with artificial intelligence (AI) requires a focus on human-centred trustworthy AI. This can be achieved by collaborations between research excellence centres with a technical focus on combining expertise in theareas of learning, optimisation and reasoning. Currently, this work is carried out by an isolated scientific community where research groups are working individually or in smaller networks. The EU-funded TAILOR project aims to bring these groups together in a single scientific network on the Foundations of Trustworthy AI, thereby reducing the fragmentation and increasing the joint AI research capacity of Europe, helping it to take the lead and advance the state-of-the-art in trustworthy AI. The four main instruments are a strategic roadmap, a basic research programme to address grand challenges, a connectivity fund for active dissemination, and network collaboration activities.

For social acceptability and for ethical purposes, it is essential to make the decisions of AI systems interpretable by humans with strong guarantees on how the data are handled. This raises scientific challenges that require complementary skills held in our research group.
Our goal is to investigate how to produce explanations for results returned by AI systems and how to build AI algorithms with guarantees of fairness and privacy, in the setting of varied tasks such as classification, recommendation, resource allocation or matching.
We will address the challenges above by combining methods from symbolic and numerical AI and by building on exploratory works done by the researchers involved in this chair.
Our results will contribute to designing more transparent and trustworthy AI systems. This will give users insights to better control, individually or collectively, the role and the use of AI systems that exploit personal data for applications directly impacting their daily life.

AI, and in particular Data Mining and Recommendation, offers the ability to observe the behavior of millions of customers and predict their future choices. While consumer choices in classical theory are based solely on preferences and on price, Behavioral Economics and in particular the study of Learning in Economics, have established that consumer behavior is largely dictated by contexts and evolves over time. Decisions are guided by heuristics influenced by psychological, emotional, cultural and social factors.
The benefits of this chair are two-fold. First, data analysis expertise will empower choice theorists by providing powerful AI tools for mining and predicting customer behavior as well as help them test their hypotheses over time. In particular, Evolutionary Economics studying how agents learn over time has yet to bridge the gap between theory and empirical testing. Second, experimental expertise will allow to focus the study on key behavioral elements, thus reducing testing costs for the Retail industry by relying on small-scale simulations in lab testing before real-life deployment.

Data growth and availability as well as data democratization have radically changed data exploration in the last 10 years. Many different data sets, generated by users, systems and sensors, are continuously being collected. These data sets contain information about scientific experiments, health, energy, education etc., and they are highly heterogeneous in nature, ranging from highly structured data in tabular form to unstructured text, images or videos. Furthermore, especially online content, is no longer the purview of large organizations. Open data repositories are made public and can benefit more types of users, from analysts exploring data sets for insight, scientists looking for patterns, to dashboard interactors and consumers looking for information. As a result, the benefit of data exploration becomes increasingly more prominent. However, the volume and complexity of data make it difficult for most users to access data in an easy way.
The core principle of INODE is that users should interact with data in a more dialectic and intuitive way similar to a dialog with a human. To achieve this principle, INODE will offer a suite of agile, fit-for-purpose and sustainable services for exploration of open data sets that help users (a) link and leverage multiple datasets, (b) access and search data using natural language, using examples and using analytics (c) get guidance from the system in understanding the data and formulating the right queries, and (d) explore data and discover new insights through visualizations.
Our service offering is formed by and will initially respond to the needs of large and diverse scientific communities brought by our three use case providers: (a) Cancer Biomarker Research – SIB Swiss Institute of Bioinformatics, Switzerland, (b) Research and Innovation Policy Making – SIRIS, Spain, and (c) Astrophysics – Max Planck Institute for Extraterrestrial Physics, Germany.

Ontology-mediated query answering (OMQA) is a recent paradigm for data access and data integration that allows to query legacy database systems by exploiting knowledge about the application domain. During the last decade, research on OMQA has essentially target the case where data comes in a relational form (or in the very close RDF format) and the paradigm is deployed on top of relational data-management systems. Yet, data generated by modern society is diverse in many terms and exhibits heterogeneity with respect to the logical models and the underlying data management systems that are actually in use. The goal of the CQFD research project is to investigate the definition and implementation of OMQA beyond the relational setting. Taking this step will give rise to novel techniques to query and analyze federated legacy data-management systems (called polystores) thereby permitting governments, companies, and institutions to gain actionable insights from their Big-Data.

Half of middle age adults exhibit more than two concomitant chronic diseases (“multimorbidity”). The progressive aggregation of chronic diseases across lifespan is drawing original health trajectories from multimorbidity to cancer. The overall goal of LIFE project is to examine the determinants of health trajectories by capturing underestimated yet crucial contributing factors including access to care, socio-economic factors, environmental exposures and urban design.
In this project, we are developing an integrated and multidisciplinary research plan uniting biological scientists, doctors, epidemiologists, public health specialists, social scientists, urban designers and big data researchers into a common framework for developing evidence-based societal and environmental control of health trajectories and transition to cancer.

Les systèmes en ligne centrés sur l’utilisateur, basés sur les données des utilisateurs, ont pris une part essentielle dans nos vies et pour notre économie. Malgré leur utilité, ils ont également apporté des menaces pour la sécurité et la vie privée qui ont pris des proportions majeures à tous les niveaux. Le but de mon projet est de fournir des méthodes et des outils pour développer des systèmes centrés sur l’utilisateur fiable qui assure la sécurité et la vie privée des utilisateurs. Je propose une approche intégrée considérant les systèmes en agrégé en combinant deux composantes clefs: (1) la vie privée: assurer le contrôle de l’information révélé e et assurer que les utilisateurs puissent comprendre comment leurs données sont manipulées; et (2) confiance: assurer que les utilisateurs puisse faire confiance aux contenus et aux personnes avec lesquelles ils interagissent anonymement.