Research on Security Assets Attention Networks for Temporal Knowledge Graph Enhanced Risk Assessment

The rapid development and extensive application of cyberspace have brought numerous opportunities to Internet users. Due to the characteristics of virtual, and open nature, cybersecurity assets are highly susceptible to attacks. Therefore, security asset risk assessment is a challenging task in the field of cyberspace security. We present Security Asset Attention Network (SeAAN), a novel model that achieve risk assessment of asset node to capture temporal knowledge graph structural evolution. Specifically, SeAAN computes risk assessment of asset node through joint attention focus on both structural neighbor and temporal history, which assigns distinct snapshots to facts at various time stamps, capturing dynamic knowledge fluctuations effectively. Extensive experiments demonstrate that SeAAN achieves significant performance on a real-world benchmark dataset for temporal knowledge graph enhanced security asset risk assessment. Moreover, our ablation analysis confirms the efficacy of integrating structural attention and temporal self-attention in a joint manner. Empirical results on real-world datasets demonstrate that our model exhibits more substantial performance enhancements compared to conventional approaches.