Dynamic entity alignment method based on attack mode

The invention discloses a dynamic entity alignment method based on an attack pattern, and relates to the technical field of network security protection, entity extraction and relation extraction are performed through a deep learning model, named entities in a text are marked by adopting a deep learning Transformer, the entities and the relation are mapped to a vector space through an embedded representation method, and CNN and RNN are used for capturing context information. The attention mechanism improves information attention, multi-task learning jointly processes entity extraction and relation extraction and then constructs a knowledge graph, extracted attack mode abstract semantic description is used as entity feature embedding, a mapped standardized format key point is used as an auxiliary entity attribute label, various relation information in a sub-graph in the knowledge graph is combined, and the attack mode abstract semantic description is obtained. And meanwhile, a time parameter set is used for inc