面向审计的严格约束的序列挖掘算法  

Audit-oriented sequence mining algorithm with strict constraints

作　　者：辛鸿亮[1] 欧阳为民[1] 祝万涛[1] 

机构地区：[1]上海大学计算机工程与科学学院,上海200072

出　　处：《计算机应用》2006年第3期630-634,共5页journal of Computer Applications

摘　　要：网络安全审计数据具有很强的时间特征。提出了面向审计基于SPAD算法的严格约束的序列挖掘快速算法(Sequence mIning with Strict Constraints,SISC),它充分利用了序列数据的时间和属性相关的特征指导挖掘,并使用严格的属性模式裁减概念等价类,提高了规则的有用度。最后在真实的审计数据集上的试验结果表明,SISC的效率优于SPADE,尤其当项的个数远大于属性的个数的时候。Security audit data has obvious time feature, but many seqnence pattern mining algorityms consider little about the time feature of sequence data. A fast algorithm SISC（ Sequence mining with Strict Constraints） was presented based on SPADE. Time and attribute-relative features were utilized to lead the mining process, and strict attribute schemes were used to prune sequential rules. Experiments on a real-world audit dataset show that SISC outperforms SPADE, especially when the number of attributes is far less than the number of attribute values.

关 键 词：数据挖掘 序列挖掘 安全审计 概念格 

分 类 号：TP311.131[自动化与计算机技术—计算机软件与理论]