We’re delighted to announce that Katja Hose, Antheas Kapenekakis, Daniele Dell’Aglio, Martin Bøgsted, and Minos Garofalakis have won the Best Paper Award at the Conference on Advances in Databases and Information Systems 2025 for their paper “Pasteur: Scaling Privacy-Aware Data Synthesis”!

Privacy-aware data synthesis aims to create synthetic data that mirrors real data without compromising privacy. While current algorithms work well for smaller datasets, they struggle with large datasets, which leads to long processing times. These processes are often confined to a single server due to privacy concerns. In this paper, the authors introduce Pasteur, a framework designed to scale data synthesis efficiently within a single server environment by using parallel processing, optimized memory usage, and a faster algorithm for calculating marginal data.

The Conference on Advances in Databases and Information Systems (ADBIS) series is a key forum for sharing research advancements and fostering collaboration among the global database, information systems, data science, artificial intelligence, and machine learning communities. These conferences have long provided an international platform for presenting and discussing innovative results in areas such as database theory, data management, and information technologies.

Congratulations to Katja Hose, Antheas Kapenekakis, Daniele Dell’Aglio, Martin Bøgsted, Minos Garofalakis on this outstanding achievement!

Abstract

Among other measures of data quality, determining the reliability of conflicting values from different sources is especially challenging. Traditional data fusion approaches often infer correct values in simple cases, but struggle to handle variations in data granularity (such as differences in temporal, spatial, or categorical aggregations) and offer limited insight into the nature of disagreements. Thus, we propose a new source evaluation approach for numerical attributes that measures discordance (i.e., the extent to which sources differ from each other). Unlike existing methods that focus solely on point estimation, we allow both fine-grained and coarse-grained analysis, allowing more sophisticated data quality assessments. We employ a linear programming solver that transparently adapts to any data alignment expressed in a set of operators resembling relational algebra. Extensive experiments on real-world datasets demonstrate that our method generalizes existing truth discovery techniques, measuring differences with Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and can adapt to diverse and complex scenarios.

About Katja Hose

Katja Hose is Professor at TU Wien Informatics, where she leads the Data Management and Knowledge-Driven AI Lab at the Research Unit Databases and Artificial Intelligence. She had prior positions at Aalborg University, the Max Planck Institute for Informatics, and received her PhD from Ilmenau University of Technology.

Her research is rooted in data management and knowledge engineering, with a particular focus on graph data management, knowledge graphs, data integration, and applied machine learning.

She has co-authored more than 150 peer-reviewed publications and serves on the editorial boards of leading journals such as the VLDB Journal, the Semantic Web Journal (SWJ), and Transactions on Graph Data and Knowledge (TGDK). In addition, she contributes to the international research community in various organizational and program committee roles at flagship conferences, including VLDB, SIGMOD, EDBT, ISWC, ESWC, and TheWebConf.

Curious about our other news? Subscribe to our news feed, calendar, or newsletter, or follow us on social media.