Practical Applications of Cyber security: A Research Paper on Knowledge Graphs for Real-World Security Solutions
Krishna Bonagiri, L. Priyadharshini, R Panneerselvi., Pinnamaraju Sahitya, Hemanth Swamy, V. P. · 2024
The realm of cybersecurity is riddled with intricate and ever-evolving cyber-attack activities, presenting formidable obstacles to cybersecurity professionals. By incorporating knowledge graphs into this field, one may help with threat identification and situational awareness by visualizing the intricate cybersecurity landscape. But creating these knowledge graphs might add noise and conflicts, which could result in incorrect conclusions. The goal of this work is to address the difficulties involved in creating and assessing cybersecurity knowledge graphs. To offer a more flexible framework appropriate for different data properties, we improved the current cybersecurity ontology and put together a cybersecurity knowledge graph dataset (CS13K). The cybersecurity knowledge graph is made with Neo4j. In addition, we developed an AttTucker model to evaluate knowledge graph quality, which is based on Transformer architecture. In order to reduce the dimensionality of knowledge embeddings and capture latent links among entities and relations, this model uses multiple self-attention heads, producing assessment results that are similar to those of higher dimensionally embedding. Additionally, in the AttTucker model-based evaluation procedure, we incorporated path-level data from knowledge graph entities. Our method significantly improves F1 value and accuracy, surpassing current models in knowledge graph quality evaluation tasks, according to an experimental investigation using cybersecurity and general datasets.