摘要
为提升Web入侵检测中聚类分析的效率和质量,提出了一种事件聚类分析方法。给出了Web日志数据的预处理办法,之后对数据模型进行定义,在分析过程中,先通过决策树预分类降低样本数据的规模,提升聚类效率,再通过替换访问用户为访问事件,引入路径离散度,改良了路径相似度的计算方法,提升聚类质量。
To increase efficiency and quality of clustering in Web intrusion detection,a new clustering algorithm is presented.Some data preprocessing methods in Web usage mining are given out,then the data model is defined.In the analytical process,firstly,large-scale sample is reduced by use Decision Tree,then the calculation method of browse similarity is improved,browse user is replaced by browse event,discrete degree is also considered.
出处
《计算机与数字工程》
2012年第2期75-78,共4页
Computer & Digital Engineering
基金
杭州科技职业技术学院科学研究课题(KZYZ-2009-3)资助
