Modelagem de sistemas de informação para a mineração de processos: características e propriedades das linguagens

Abstract

Storing information in large data repositories (Big Data) creates opportunities for Organizations to use Process Mining techniques to extract knowledge about the performance and actual flow of their processes of business. One of the fundamental elements for achieving this objective is the relationship between process modeling languages, process event logging (logs) and Process Mining algorithms. In this work, comparisons were made between three languages (BPMN, Petri Nets and YAWL) which are usually used to model business processes with respect to their capabilities of use in Process Mining, especially in Process Discovery. The models created were based on typical Workflow patterns and five scenarios were simulated for each language using three Process Discovery algorithms (Alpha, Heuristic Miner and ILP Miner). The results indicate that the choice of the language used in the modeling and in recording of the business processes influences the quality of the results obtained by the Process Discovery algorithms. This work also presents suggestions for the development of process modeling languages and process mining algorithms.