序列模式挖掘在网络告警分析中的应用

An Application of Sequential Pattern Mining in Network Alarm Data Analyses

序列模式挖掘可以用来有效地发现网络系统中的告警关联知识.论文研究了序列模式挖掘在网络告警分析中的具体应用.首先,将挖掘过程分成了特定设备告警序列挖掘、同类设备告警序列挖掘和互联设备告警序列挖掘等3类,根据不同的用户意图来有效地确定挖掘范围,避免对无关数据的访问.为了进一步提高挖掘算法的执行效率,又提出了用于描述网络拓扑信息的拓扑约束,并设计了基于拓扑约束的互联设备告警序列模式挖掘算法.

Sequential pattern mining provides an effective way to find the knowledge of network alarms. Applications of alarm sequential pattern mining are studied in this paper. Firstly, the mining procedures are categorized into three kinds such as alarm sequential pattern mining of special device, device type and connected devices to focus on the alarm data based on users' intentions. Further, a mining constraint called topological constraint is proposed, which extends a new way to describe the network topological information. And an algorithm for alarm sequential pattern mining of connected devices is designed based on topological constraints.