advanCed pRivacY Preserving TechnOlogies for enterprise knowledge GRAPHs and Artificial Intelligence

This project initiates the development of CRYPTO4GRAPH-AI, a scalable, secure, and privacy-preserving framework for intelligent management and analytics of graph data in and across enterprises and other organizations. Knowledge Graphs (KGs) have gained increasing adaptation and popularity in enterprise settings, due to their ability to integrate data, rich domain knowledge, and metadata from heterogeneous sources. Despite considerable research effort on privacy-preserving analytics with standard database technologies in recent years, developing privacy notions and secure analytic solutions for graph data is relatively unexplored. The main innovation of CRYPTO4GRAPH-AI is the way of enabling secure and efficient analytics on graph data by employing state-of-the-art semantic technologies, machine learning (ML), and cryptography. To the end, CRYPTO4GRAPH-AI will leverage ML solutions, considering the aspects like privacy of data owners, confidential data preservation, data sovereignty, efficient knowledge management, as well as the scalability, accuracy, and federated analytics for graph data. By three core contributions, CRYPTO4GRAPH-AI will go beyond existing approaches: (1) privacy-preserving tools to perform ML on enterprise graph data; (2) privacy-preserving analytics on graph data by improving state-of-the-art algorithms.; (3) demonstrating the feasibility of the CRYPTO4GRAPH-AI framework and its extensions over different use cases in the domains of cybersecurity and supply chain. The CRYPTO4GRAPH-AI consortium includes two German partners who are experts in semantic and ML technologies (Research institute: Fraunhofer FIT, Company: eccenca GmbH) and two French partners specialized in "Privacy Enhancing Technologies'' (Research institute: INRIA, Company: Cosmian). CRYPTO4GRAPH-AI will facilitate knowledge exchange between the two countries and conduct interdisciplinary research to develop new solutions to improve privacy for KGs.