چكيده به لاتين
Abstract
The business operations of today’s enterprises are heavily influenced by numerous number of internal and external business events. On the other hand, by using Complex Event Processing (CEP), complex patterns of events and therefore specific situations could be identified. Business Activity Monitoring (BAM) applications are capable of online analysis of events related to business processes using CEP. In Process-Aware organizations, a typical process model describes organization’s expected workflow but due to numerous restrictions, this normative procedure could not be constrained, therefore it would be likely that real world business instances deviate from normative procedure (described in process model). A part of BAM System’s tasks, as Conformance Checking, covers reporting this kind of deviations. Numerous mechanisms has been proposed to report deviations from model by generating CEP rules as Anti-Patterns from Casual Behavioral Profiles which derived from business process model. In industry use cases, with complex process models having multiple instances running simultaneously, such mechanisms dealing with high rate of input events and CEP rules, therefore suffer from consuming lots of computing resources. In response to these challenges, in this thesis, a distributed, scalable mechanism proposed which by dispatching monitoring process of process instances among CEP nodes balances the load among processing nodes. Also, a method proposed that enhanced keeping partial matching instances in memory which leads to significant saving of computing resources. The distributed mechanism has been implemented and its load balancing and scalability approved by accomplished experiments. The proposed enhancement in conformance checking rules has been implemented and according to accomplished experiments, for different rates of process instances, has better performance comparing to other event driven conformance checking mechanisms.
Keywords: Business Activity Monitoring, Complex Event Processing, Business Process Management.
