PrivyKG: Security and Privacy Preservation of Knowledge Graphs Using BlockChain Technology

Abstract

Recently, knowledge graph has proved its benefits in AI applications and especially in knowledge representation and reasoning. Therefore, with the huge amount of published data, knowledge graph privacy, trust and integrity take an important role to protect it from illegal access and modification. In this paper, we propose PrivyKG, a new framework which uses the blockchain technology along with an access control based on it to preserve the knowledge graph privacy, integrity, and ensure security and trust. The proposed work has two essential parts to build a decentralized knowledge graph, where the first one uses off-chain storage for ensuring the blockchain scalability and the second one integrates a permissioned Hyperledger Fabric powered with GO smart contracts. The proposed framework is evaluated with a DBpedia dataset that illustrates the efficiency and the feasibility of our proposed against potential threats. Povzetek: Predstavljena je metoda decentraliziranih grafov znanja za bločne aplikacije.