A General Construction Method of Cyber Security Knowledge Graph

The corpora in cybersecurity knowledge base is characterized by multi-entity and weak relation. In order to solve the problem that entity extraction, entity links and entity disambiguation can lead to wrong knowledge, a knowledge graph construction method for security knowledge base is proposed. At the conceptual level, the cybersecurity ontology model and ontology-instance model are constructed based on the five-tuple model structure, and the ontology-instance model is classified and associated from the horizontal and vertical perspectives. In the data layer, the ontology features are combined with the named entity recognition model. While recognizing entities, the ontology-instance model is optimized. By comparing with other entity recognition models, the accuracy and efficiency of the optimized ontology-instance model are verified.