Keynote Lectures

Location Privacy in Crowdsourcing
Richard Chbeir, Université de Pau et des Pays de l'Adour (UPPA), France

An Overview of AI & Law Models of Case-Based Reasoning, with Applications in Decision Analysis and Explainable AI
Henry Prakken, Utrecht University, Netherlands

Data Quality and Ethics in Cultural Heritage Knowledge Graphs
Valentina Presutti, University of Bologna, Italy

Abstract
The rise of mobile technologies and spatial crowdsourcing platforms has led to a proliferation of geolocated data, widely exploited for applications such as traffic management, urban planning, and task assignment. However, this wealth of information brings significant privacy risks for users, particularly due to inference attacks, which aim to extract sensitive information from recurring spatio-temporal patterns. For example, analyzing a user's regular movements can reveal their home address, workplace, or personal habits such as medical visits, religious practices, or union-related activities. Even after anonymization, cross-referencing datasets is often sufficient to reconstruct an accurate user profile. In this talk, we will explore how traditional privacy-preserving techniques (such as pointwise anonymization, noise addition, or dummy location generation) often prove insufficient. Two main angles will be particularly explored: 1) Data representation and how the identification of Points of Interest (POIs) in considered so far, and 2) Anonymization mechanisms and how existing methods generate fictitious locations. Some results of our research groups will also be presented.

Abstract
In this talk I will give an overview of recent research of myself and my students on formal and computational models of legal case-based reasoning. Legal case-based reasoning is the process of arguing for or against decisions in new cases by drawing analogies to or stressing differences with precedent cases. In the field of AI & law, seminal work on case-based reasoning was by Ashley & Rissland on the HYPO system, followed by Aleven's work on the CATO system. This work primarily focussed on generating case-based debates. More recently, John Horty initiated the formal study of so-called precedential constraint, which addresses a question of a more logical nature, namely, to what extent a decision in a new case is constrained by a body of precedents.

In our recent work we have further developed this work and, among other things, developed gradual consistency measures for collections of case-based decisions. We have also studied the application of models of legal case-based reasoning in explainable AI, by exploiting an analogy between the legal notion of precedent and the machine-learning concept of training data.

Abstract
Digital cultural heritage risks turning into a distorted echo of the past, built on fragmented and biased data. This keynote argues that ensuring data quality is not merely a technical challenge, but an ethical responsibility: Knowledge Graphs (KGs) that represent our heritage actively shape historical narratives and influence the reliability of AI systems. The talk will highlight both the technical and ethical dimensions of this issue, focusing on challenges such as data heterogeneity and cultural bias. Drawing on examples from our research such as ArCo, the KG of Italian Cultural Heritage, and ChoCo, a KG integrating musical harmony datasets, I will illustrate practical methodologies to enhance data quality, while also exposing their limits and pointing to open research challenges that demand attention.