摘要
为细粒度分析多维度组织网络空间中威胁情报,提出一种结合双向长短时记忆网络Bi-LSTM与线性链条件随机场CRF的实体识别模型。利用网络空间安全领域词典构建、词向量训练、序列标注以及模型训练方法建立了知识图谱,通过Bi-LSTM提取特征识别网络空间安全领域中12类命名实体。结果表明,该方法评价值优于其他算法,F值达到85.00%,整体识别性能较高。
This paper proposes an entity recognition model combining Bi-LSTM and CRF of linear chain to analyze the threat intelligence in multi-dimensional cyberspace with fine granularity.The study involves establishing the knowledge graph using domain dictionary construction,word vector training,sequence labeling and model training in the cyberspace security domain,and recognizing 12 types of named entity recognition in cyberspace security domain by Bi-LSTM feature extraction.The results show that this method boasts the evaluation value superior to other algorithms,with the F value of up to 85.00%,and thus a higher overall recognition performance.
作者
廉龙颖
Lian Longying(School of Computer & Information Engineering, Heilongjiang University of Science & Technology, Harbin 150022, China)
出处
《黑龙江科技大学学报》
CAS
2020年第6期717-722,共6页
Journal of Heilongjiang University of Science And Technology
基金
黑龙江省自然科学基金项目(F201436)。
