Abstract: I will discuss some of our experiences in Luxembourg on mining and reasoning with legal texts (MIREL). I will focus on three challenges we are investigating since a decade. First, the challenge of mining norms and arguments in legal texts. Though deontic logic and normative systems have been studied for at least three decades in computer science, and much longer in other disciplines, only recently argument and norm mining tools have become available in natural language processing. Second, for the integration of mining and reasoning in the MIREL pipeline, legal interpretation turned out to be a key challenge, and legal ontologies an important part of the solution. Third, for the challenge of automated legal decision support - and especially decision making - , we are working towards deontic explanations using language generation tools.

Bio: Leon van der Torre is a full professor of artificial intelligence in Luxembourg since 2006. With various co-authors he introduced input/output logic, the BOID agent architecture, the game-theoretic approach to normative multiagent systems, the ArchiMate approach to enterprise architecture, the Jiminy architecture for AI ethics, and the LogiKEy framework and methodology for legal and ethical reasoners. He is founding member of the European doctoral school on Law, Science and Technology (LASTJD, 2012-) and he was the PI of the FP7 MC RISE project MIREL (2016-2019).  He is editor of handbooks on deontic logic and normative systems, formal argumentation, normative multiagent systems, and legal AI. He is deputy editor in chief of the Journal of Logic and Computation and member of the editorial board of Journals of Applied Logics, Argument & Computation, and AI & Law.