Workflow Mining: A new approach using Genetic Algorithm

La presente tesi rappresenta un lavoro svolto in Danimarca come esame finale del corso di Laure in informatica.

In this thesis a quite new technology called workflow will be described, what it is and how it works. I will give an overview of the system architecture within a business environment, the basic workflow lifecycle and the basic related concepts. Then I will reach the modelling issue, giving common terminology and trying to make the reader understand deeper how to model and the workflow routing possibilities. Once the modelling part is reached it will be obvious for a designer to know if the model already designed is well done, portable, understandable for others, and if the model does what he expects it to do. There has been introduced a very new possibility of using mining technologies to discover models from workflow log that will be used not only for the so called “delta analysis”, but also to “rediscover” new models, giving an alternative, where it is possible to the already designed one or giving some interesting tips. The rediscovering process touches not only the control flow analysis, but also the resources analysis and the workload analysis. In this thesis the control flow analysis is considered, before a Heuristic approach is used to deal with the rediscovering process, then, using genetic algorithms to rediscover the common control flow possibilities. A new approach will be introduced, the coevolutionary algorithm that aims to improve the already known genetic algorithm making an evaluation analysis between them.