Discovering actionable patterns in event data

摘要: Applications such as those for systems management and intrusion detection employ an automated real-time operation system in which sensor data are collected processed real time. Although a effectively reduces the need staff, it requires constructing maintaining correlation rules. Currently, rule construction experts to identify problem patterns, process that is time-consuming error-prone. In this paper, we propose reducing burden by mining historical readily available. Specifically, first present efficient algorithms mine three types of important patterns from event data: bursts, periodic mutually dependent patterns. We then discuss framework efficiently events have multiple attributes. Last, Event Correlation Constructor--a tool validates extends knowledge.