Keeping the Privacy and the Security of the Knowledge Graph Completion Using Blockchain Technology

Abstract

Nowadays, Knowledge Graph becomes very important for representing and reasoning about knowledge data. Its name was known from Google knowledge graphs in 2012. Many IA applications have used it for managing the various types of knowledge data and especially the personal data. Therefore, ensuring security, trust, integrity and privacy of its content was the goal of the AI researchers. KG completion is one of the most methods that its aim is completing the KG with missing true triples. In this paper, we propose to use blockchain technology to keep the security and privacy of the KG completion based on KG embedding that embed relations and entities in continues vectors spaces. The completion tasks can benefit from the decentralization feature of BC. Our design proposes to use an offchain storage in order to improve the scalability of BC network that contains only critical KG embedding data. The implementation and the evaluation have used Hyperledger Fabric (HF), MangoDB and several KG completion datasets.