基于改进BiLSTM-CRF模型的网络安全知识图谱构建

Knowledge graph construction for network security base on modified BiLSTM-CRF

针对网络安全领域的图谱构建任务,基于BiLSTM-CRF模型引入了外部网络安全词典来加强网络安全文本的特征,并结合多头注意力机制提取多层特征,最终在网络安全数据集取得了更优异的结果。利用企业内部的日常网络运维数据,设计并构建了一个面向企业网络安全运维管理的知识图谱,为后续进一步研究基于图谱的企业网络安全智能决策等应用奠定基础。

On focus of the task of constructing graphs in the field of network security,an external network security dictionary is introduced based on the BiLSTM-CRF model to enhance the features of network security texts,and a multi head attention mechanism is combined to extract multi-layer features.Better results are achieved in the network security dataset.A knowledge graph for enterprise network security operation and maintenance management is designed and constructed by means of daily network operation and maintenance data within the enterprise,laying the foundation for further research on intelligent decision-making of enterprise network security based on graphs.