【作者】 余峰； 汤庸； 潘炎；

【Author】 Yu Feng,Tang Yong,and Pan Yan (College of Information Science and Technology,Sun Yat-sen University,Guangzhou 510275)

【机构】 中山大学信息科学与技术学院；

【摘要】 近年来,工作流挖掘越来越受到工作流管理的研究者们的青睐.一些企业没有工作流系统,且企业业务流程会随时间动态变化.考虑到手工建立工作流模型是困难且耗时的工作,工作流管理者需要有效的方法来自动建立和定时更新工作流模型.为此,提出了一种利用事件日志挖掘工作流模型的算法,可以学习带有循环结构的工作流模型.接着用一个具体的实例来说明算法的执行过程.最后,通过实验比较来说明该算法可以利用活动的时间区间的重叠来排除活动间的依赖关系,明显减少了算法中时间昂贵的独立性测试的次数,从而改善算法性能.更多还原

【Abstract】 Recently,workflow mining has attracted more and more researchers’ eyes in workflow management.Some organizations do not have workflow systems.And business processes evolves dynamically with time.With the consideration that creating a workflow model by hand is complicated and time-consuming,a learning algorithm is proposed to automatically discover workflow models with iterative structures.And an example is given to illustrate the execution process of the algorithm.Finally, experiment results show that the algorithm uses the overlap of activities’ intervals to eliminate some dependency relations of activities,so as to reduce the number of time-consuming independence tests and improve the algorithm’s performance.更多还原