Scalable Alignment of Process Models and Event Logs: An Approach Based on Automata and S-Components
Given a model of the expected behavior of a business process and an event log recording its observed behavior, the problem of business process conformance checking is that of identifying and describing the differences between the model and the log. A desirable feature of a conformance checking technique is to identify a minimal yet complete set of differences. Existing conformance checking techniques that fulfil this property exhibit limited scalability when confronted to large and complex models and logs. This paper presents two complementary techniques to address these shortcomings. The first technique transforms the model and log into two automata. These automata are compared using an error-correcting synchronized product, computed via an A* that guarantees the resulting automaton captures all differences with a minimal amount of error corrections. The synchronized product is used to extract minimal-length alignments between each trace of the log and the closest corresponding trace of the model. A limitation of the first technique is that as the level of concurrency in the model increases, the size of the automaton of the model grows exponentially, thus hampering scalability. To address this limitation, the paper proposes a second technique wherein the process model is first decomposed into a set of automata, known as S-components, such that the product of these automata is equal to the automaton of the whole process model. An error-correcting product is computed for each S-component separately and the resulting automata are recomposed into a single product automaton capturing all differences without minimality guarantees. An empirical evaluation shows that the proposed techniques outperform state-of-the-art baselines in terms of computational efficiency. Moreover, the decomposition-based technique is optimal for the vast majority of datasets and quasi-optimal for the remaining ones.