Methods of discreet mathematics, mathematical logic, theory of algorithms, theory of automata and Petri nets, linear algebra, software engineering and simulation were used in the course of project work.
Empirical base of research
The empirical base for the study was based on event logs, obtained from information systems, operating in various application domains, such as banking, insurance, healthcare, manufacturing, municipalities, and online ticketing. These event logs were used to verify both developed and existing process mining approaches. Artificial event logs produced by specially developed log generators were used as well.
Results of research
Research was carried out in two interrelated directions. The first is the modeling and analysis of complex information systems. The second is related to process mining and analysis of systems behavior based studying of event logs generated by information systems.
In the field of modeling and analysis of complex information systems the following theoretical and practical results were obtained.
- A new version of algorithm for constructing the reachability graph for time Petri nets has been developed. It was shown that the algorithm applied to a special revealed class of time Petri net is more effective.
- A novel method of verification of nested Petri nets, based on constructing of net unfoldings has been developed. An approach for the construction of component-based unfolding was suggested and the correctness of this approach was proven.
- An approach for the analysis of behavioral properties of Petri nets, based on transition priorities, has been developed. An algorithm for the construction of live and bounded Petri nets from live and unbounded Petri nets, using transition priorities, has been proposed. The correctness of the algorithm was proven.
Also in the field of process mining new theoretical and practical results were obtained.
- A new method of mining structured process models from decomposed transition systems has been developed.
- ProM tool plug-ins for mining and analysis of BPMN process models have been implemented.
- A new algorithm for conformance checking between high-level models and low-level event logs has been developed. The correctness of the algorithm in the case of the perfect fitness was proven. The algorithm is implemented in ProM.
- Methods for generating event logs (either containing noise or not) have been developed and implemented as ProM plug-ins. A comprehensive analysis of produced logs was performed.
- An application of process mining techniques to the software processes analysis was examined. A novel Software process mining research direction was established; some new approaches in this direction were proposed.
- DPMine,a language for modeling and conducting experiments, has been improved and implemented in the form of individual DPMine/C library. VTMine modeling toolbox has been developed and partially implemented as individual software. DPMine/C has been integrated into VTMine as a set of plug-ins.
The results obtained during the work on the project can be used for further research in the field of process-aware information systems as well as for process modeling, analysis and verification performed for real information systems in order to improve their reliability and productivity.
Level of implementation, recommendations on implementation or outcomes of the implementation of the results
A variety of software tools and toolboxes is the main practical result obtained in the course of project work. Particularly, the following plug-ins for ProM toolbox have been implemented: “Discovery using TS decomposition”, “Analyze BPMN diagram”, “Gena: Event Log Generator”. The plugins are available in the ProM packages repository. They can be used by ProM users to analyze their event logs.