Abstract: Complex event processing （CEP） needs to correlate multiple continuous events involved in complex temporal relationship and attribute logic relationship to more abstract complex events in richer semantic through rule matching and reasoning， then trigger the down-stream actions. Event query language is to specify event pattern， and the event algebra underlying determines expressiveness of event query language， and also indirectly affects performance of event detection. Thus， the in-depth analysis of events algebraic laws is one of very important approaches to trade-off the event detection performance and expressiveness of query language. Based on White event algebra， the formal framework is extended for event algebra， considering semantic definition of events in more concise temporal relationships of events to enhance the expressiveness of event query language while taking into consideration event selection and consumption policy-the context consumption policy to improve the performance of event detection. The algebraic laws of InforSIB event algebra and its context consumption policy are discussed. Context consumption policy is very suitable for real-time monitoring applications and has good algebraic laws to ensure that the time and space complexity is bounded for composite event detection.

Key words: InforSIB event algebra, context consumption policy, event consumption policy, algebraic laws

摘要： 复合事件处理系统多节点处以及系统外部生成的大量的、连续到达的事件，通过规则匹配、推理等方法对事件模式进行检测，连续输出经过组合后的复合事件，以触发相应的后继处理。事件查询语言是规约事件模式的声明性语言，事件代数决定了事件查询语言的表达能力，同时也间接影响了事件检测的性能。因此，在权衡检测性能与事件查询语言的表达能力时，深入分析事件代数是非常重要的手段。基于White事件代数，扩展了事件代数的形式化框架，在形式化框架中考虑复合事件在更精细的时间关系下的语义定义，以增强事件代数的表达能力，同时考虑与应用相关的事件选择和消费策略-上下文策略，以提高事件检测的性能。讨论了InforSIB事件代数的代数性质，提出的上下文策略非常适合用于实时监控等领域，且具有良好的代数性质，保证了复合事件检测的时间和空间复杂度是有界的。

关键词: InforSIB事件代数, 上下文策略, 事件消费策略, 代数性质