导　　师： 林树宽

学科专业： 081203

授予学位： 硕士

作　　者： ;

机构地区： 东北大学

摘　　要： 事件流作为数据库领域的新兴热点问题,近年来获得了日益广泛的关注。很多应用在本质上都是事件驱动的,例如,RFID采集到的元数据可以看做是一种简单事件,商业事务处理中每次交易事务或数据变化也可以作为事件,此外还包括网络报警事件监测、传感器网络监测等等。在面向实时监控的应用中,这些连续到达的事件形成了语义丰富的无界事件流。 现代业务流程随时都在产生海量的、不同种类的事件,要想从中发现有价值的事件无疑是大海捞针。帮助企业发现感兴趣或有意义的事件,需要一种新的软件模式来处理数量众多的不同事件,逐步兴起的事件检测技术无疑起着举足轻重的作用。 为了能从实时到达的事件流的海量事件中进行有效复杂事件检测,往往需要考虑事件流的更新淘汰问题,事件流中乱序事件处理,非事件处理以及查询结果的优化。基于事件检测中以上常见问题,本文主要工作如下： /(1/)首先描述了一个RFID事件流应用场景,针对该场景设计了哈希链表复杂事件检测算法,算法利用哈希链表特殊结构来保存中间结果,为了加快匹配过程,设计了有效的哈希映射。 /(2/)针对事件流可能出现的乱序问题,利用哈希链表特殊的结构进行了有效的处理。 /(3/)由于事件流的实时性和无限性,需要及时清理内存中过期的中间结果,本文利用滑动窗口约束,提出一种有效的事件流更新淘汰机制。 /(4/)事件检测往往是为了进行有效的事件预测,为了避免重复预测,引出最小发生事件概念,并利用规则树来有效查找最小发生事件。 实验和分析证明,本文提出的复杂事件事件流检测方法具有理论上的可行性和操作上的高效与正确性。 Event stream management has become a very popular research issue in database field and gained wide attention in recent years. Actually, lots of applications are event-driven. For example, the metadata collected by RFID equipments are simple events. The transaction and business data changes also can be considered as events in businiess process management. In these application areas, these continuous arrival events constitute event streams without bound. Modern business processes produce massive and different types of events. Therefor, it's very difficult to find the valuable event from these events. To help enterprises find interesting or significant events, we need a new software model to handle a large number of different events. The rising event detection technology undoubtedly plays an important role. In order to detect the complex event from huge volume of event streams effectively, we often need to consider updating and deleting of event stream, processing over out of order event stream, negation event processing and optimization of query results. Based on the above-mentioned problems in event detection, the contributions of the thesis are as follows: /(1/) An RFID event flow scenarios is described first and the hash list complex event detection algorithm is presented based on it. Intermediate results are stored in the special structure of the hash list, and design an effective hash map in order to speed up the matching process. /(2/) For out of order events, this paper deals well with them by using the special structure of the hash list. /(3/) Due to event stream has characteristics of real-time and infiniteness. cleaning up the outdated intermediate results in memory timely is required. Here, this paper proposes an effective event stream updating and deleting mechanism based on sliding window constraints. /(4/) Event detection is often for the purpose of event prediction effectively. In order to avoid duplicate predictions, this paper presents the definition of the minimal occur

关 键 词： 事件流 事件检测 复杂事件处理 最小发生事件 情节规则树

领　　域： [自动化与计算机技术] [自动化与计算机技术]