Blockchain Threat Intelligence Knowledge Graph Alignment Via Graph Convolutional Networks

The escalating prevalence of security incidents in the blockchain sphere is posing sig- nificant challenges to its future development. The integration of knowledge graphs into blockchain security is being investigated as a potential solution to offer a com- prehensive view of the blockchain security landscape. Despite the promise, the di- versity and subpar quality of existing blockchain threat intelligence data complicate the use of knowledge graphs for representing this information. The paper proposes the use of knowledge graph fusion, particularly focusing on entity alignment and en- tity linking, as an innovative approach to reconcile knowledge graphs of blockchain threat intelligence from disparate sources. Additionally, it utilizes GCN to model the structural information and an improved TransE to model the attribute information. By combining both representations, the accuracy of blockchain threat intelligence knowledge graph alignment is significantly improved.