面向网络管理知识获取的一种序列模式挖掘新算法

为了从反映网络状况和网络行为的海量历史数据中提取有用的网络管理知识,提出了一种用于获取网络管理知识的序列模式挖掘新算法--基于定位索引投影的序列模式挖掘算法.该算法采用模式增长、分而治之的思想,通过引入定位索引的概念,避免了对投影数据库的扫描,进一步减少了系统的开销.实验证明,利用该算法能够有效地发现网络告警事件序列模式.

Novel sequential patterns mining algorithm for network management knowledge acquirement

In order to ext ract netw o rk management know ledg e f rom immense historical data, which ref lect s the state and the behaviors of the netwo rk, a nov el sequential pat terns mining alg orithm named locat ion index- pro jected sequent ial pat tern mining ( LiSpan) w as propo sed. Based on the pat ter n-g row th and divideand- rule thoug ht , this algor ithm avoided scanning the pr ojected database and reduced the sy stem load by using the locatio n index o f the items in the sequent ial database. The exper imental result s indicate that the netw ork w arning event sequent ial patterns can be discovered by using LiSpan.