Home // DBKDA 2025, The Seventeenth International Conference on Advances in Databases, Knowledge, and Data Applications // View article

Privacy-preserving Data Sharing Collaborations: Architectural Solutions and Trade-off Analysis

Authors:
Michiel Willocx
Vincent Reniers
Dimitri Van Landuyt
Bert Lagaisse
Wouter Joosen,
Vincent Naessens

Keywords: privacy enhancing technologies, data collaborations, anonymity, utility

Abstract:
Businesses and governments possess vast data with potential for analytical insights in areas like business intelligence (consumer behavior, business solvability) and governmental insights on population (crime, fraud). However, two primary challenges hinder the adoption of data-driven analytics: the lack of in-house expertise and the absence of sufficient data, which often requires collaboration with third parties. Such partnerships, especially involving ML (Machine Learning), raise concerns due to the sensitive nature of the data. This paper outlines two realistic use cases and proposes two privacy-preserving data sharing architectures tailored for business-to-business and government-to-business contexts. The first architecture uses de-identification techniques before and during data transmission, while the second assumes an already existing baseline ML model to test and refine predictions without sharing data. We present an in-depth analysis and evaluation of these architectures focusing on their complexity, trust requirements, and data-sharing efficacy.

Pages: 32 to 41

Copyright: Copyright (c) IARIA, 2025

Publication date: March 9, 2025

Published in: conference

ISSN: 2308-4332

ISBN: 978-1-68558-244-9

Location: Lisbon, Portugal

Dates: from March 9, 2025 to March 13, 2025