一种基于部件CNN的网络安全命名实体识别方法

基于知识图谱的网络安全动态预警方法,能够主动感知和应对网络安全攻击,增强感知的实时性和精准性。然而,在构建网络安全知识图谱的实体抽取过程中,传统的命名实体识别工具和方法无法识别网络安全领域中的特定类别实体,文本中的未登录和中英文混合的网络安全实体也难以被准确识别。网络安全文本中的网络安全命名实体存在中英文混合、单词缩写等问题,仅基于字的命名实体识别方法难以充分表征字或词的语义信息。因此,论文考虑中英文更细粒度的部件语义捕捉字或词的语义特征,提出一种基于部件CNN的网络安全命名实体识别方法(C C-NS-NER),利用部件CNN抽取词语部件特征中的关键语义特征,丰富字词级别的语义信息,并引入BiLSTM-CRF确保抽取字向量和部件特征中的抽象信息,同时获取标签之间的关联信息,识别文本中的网络安全命名实体。在人工标注的网络安全数据集上的实验结果表明,该方法相较于传统模型,能有效获取字或词的部件语义信息,显著提高网络安全命名实体识别的效果。

A Network Security Named Entity Recognition Method Based on the Component CNN

A network security dynamic early warning method based on knowledge graph can actively detect and respond to net work security attacks,and enhance the real-time and accuracy of perception.However,in the process of constructing the entity ex traction of the network security knowledge graph,the traditional named entity identification tools and methods cannot recognize spe cific categories of entities in the network security domain,and the network security entities that are not logged in and mixed in Chi nese and English are difficult to be accurately identified.The network security named entities in the network security text have prob lems such as Chinese-English mixture and word abbreviation.It is difficult to fully characterize the semantic information of words or words based on the word-based named entity recognition method.Therefore,this paper considers the semantic features of the Chi nese and English fine-grained parts semantics to capture words,and proposes a CNN-based network security named entity recogni tion method(CC-NS-NER),which uses the CNN to extract the features of the component word parts'key semantic features,rich semantic information at the word level,and the introduction of BiLSTM-CRF to ensure abstract information extracted from word vec tors and component features,and to obtain association information between tags and identify network security named entities in the text.The experimental results on the manually labeled network security dataset show that compared with the traditional model,this method can effectively obtain the semantic information of the words or words,and significantly improve the recognition of network se curity named entities.