【作者】 刘红；

【导师】 许志才； 方贤文；

【作者基本信息】 安徽理工大学 ， 应用数学， 2017， 硕士

【摘要】 业务流程管理有很多优点,比如:提高企业运行效率,尽可能的缩短运行时间,通过灵活的运用业务流程管理解决各种变化问题。同一个流程可能对应不同的模型,但有的不符合要求,因此会产生模型变化。同时,业务流程的质量取决于模型的优劣,而业务流程挖掘伴随互联网时代快速发展,它是一种管理领域的较新应用,分析业务流程对应的日志,挖掘可能存在的隐变迁、阻塞变迁,获得更加完备的流程模型,同时提高模型的工作效率。所以结合配置变迁优化模型是目前企业管理的重点。目前对给定目标模型的变化域的分析大部分通过变化的活动节点搜索变化范围,计算过程繁琐;挖掘业务流程模型隐变迁主要基于业务流程序列进行分析,但很少考虑跨序列间的关系,因此对挖掘业务流程隐变迁有一定的影响;仅仅给定日志,没有目标模型时,常常通过静态定位变化区域,然后优化分析业务流程模型,完善已有的模型。本文结合Petri网的方法和行为轮廓的思想,通过流程树切挖掘业务流程模型隐变迁,以及考虑在给定目标模型或者没有目标模型的情况下,从不同角度对变化域进行优化分析。本文的主要贡献如下:针对给定目标模型,寻找目标模型变化域的问题,提出基于行为轮廓的交互行为,通过边界变迁和跨边界变迁进行优化,确定所有的可疑点,通过服从度和支持率确定目标模型的变化域,不仅简化计算,而且减少了分析变化域的难度。针对挖掘业务流程模型隐变迁,并进行优化的问题,给出流程树切挖掘业务流程模型隐变迁的方法。首先根据发生频数较高的日志序列得到初始模型,然后再根据流程树切预处理事件日志,找到可能变化的区域,挖掘可能存在的隐变迁,最后通过评价指标判定带隐变迁的模型是最优模型,说明模型中存在隐变迁。该方法不仅从业务流程序列进行分析,而且考虑跨序列间的关系,从多个角度对模型进行分析,克服了已有方法的局限性。对于业务流程模型的智能优化问题,结合Petri网的方法和行为轮廓的思想,从日志行为方面分析检测业务流程模型的变化域,首先利用日志与模型的匹配关系以及动态定位来确定模型中的变化域,然后考虑接口部分,通过配置变迁优化模型,最后利用行为轮廓一致性度分析判定配置后模型是最优模型,从而克服静态寻找变化区域的有限性。更多还原

【Abstract】 Business process management has many advantages,such as improving enterprise operational efficiency,shortening the running time as soon as possible,the flexible use of business process management to solve all kinds of change problems.The same process may correspond to different models,but some do not meet the requirements,it will produce changed model.At the same time,the quality of the business process depends on the pros and cons of the model.And with the rapid development of the Internet era,business process mining has been a new application field of management,the analysis of business process correspond to the event log,mining hidden changes or blocking changes to obtain a more complete process model,and improving the work efficiency.Currently,enterprise management has focused on combining with the configurable change to optimize model.Now,for a given target model,the analysis of the changes range mainly through the active node,the calculation is complicated;mining hidden changes of business process model mainly based on sequence analysis,but rarely considering the inter relationship between sequences,so it has a certain impact on the mining business process hidden change;for given the log,but no target model,often through static positioning change region,and then optimizing and analyzing the business process model,improving the existing model.In this paper,combining with the behavior profile and the Petri net,mining hidden changes of business process model through the process tree cut and considering the given target model or no target model,optimizing and analyzing domain changes from different angles.The main contributions are as follows:For a given target model,to search domain change problem of model,optimizing boundary transitions and inter-boundary transition based on behavioral profiles interactive behavior to determine uncertain range.And combining with support and compliance to find target model change region,not only simplifying the calculation,but also reducing the difficulty of analyze change range.According to mining and optimizing hidden changes of business process model,the method to find hide transitions of business process is proposed based on the process tree cut.Firstly,the source model is obtained according to the higher frequency of occurrence log sequence.Then preprocessing log by the process tree cut,comparing the log relationship with the relationship of the initial model,and finding changes region,mining the possible hide transitions.Determining the optimal model with hide transitions through the evaluation criteria.This method which analyzes business process model not only from the business process sequence,but also consider the inter relationship between sequences and from multiple perspectives,to overcome the limitations of existing methods.The intelligent optimization of business process model,combining the method of the Petri net and behavior profile.Firstly,from the log behavior angle to dynamic analysis change parts of the business process,matching with the relation between log and model,and dynamic localization,determining the change region of model,then considering the interface part,finally using the fitness and behavioral appropriateness,optimizing model with the configuration change and it is the best model,to overcome the limitation of static search change range.更多还原