Semantic fuzzy mining: Enhancement of process models and event logs analysis from syntactic to conceptual level

Abstract

Semantic-based process mining is a useful technique towards improving information values of process models and analysis by means of conceptualization. The conceptual system of analysis allows the meaning of process elements to be enhanced through the use of property characteristics and classificati on of discoverable entities, to generate inference knowledge that can be used to determine useful patterns and predict future outcomes. The work in this paper presents a Semantic-Fuzzy mining approach that makes use of labels within event log about real-time process to provide a method which allows for mining and improved process analysis of the resulting process models through semantic – annotation, representation and reasoning. Qualitatively, the study shows by using a case study of Learning Process – how data from various process domains can be extracted, semantically prepared, and transformed into mining executable formats to support the discovery, monitoring and enhancement of real-time domain processes through further semantic analysis of the discovered models. Also, the paper quantitatively assess the level of accuracy of the classification results to predict behaviours of unobserved instances within the process knowledge-base by determing which traces are fitting or not fitting the discovered model by using a training set and test log for the cross-validation experiment. Accordingly, the work looks at the sophistication of the proposed semantic-based approach and the discovered models, validation of the classification results and their influence compared to other existing benchmark techniques and algorithms for process mining. The experimental results and data validation ends with the supposition that a system which is formally encoded with semantic labelling (annotation), semantic representation (ontology) and semantic reasoning (reasoner) has the capability to lift process mining analysis and outcomes from the syntactic level to a much more conceptual level, resulting in a mining approach that is able to induce new knowledge based on previously unobserved behaviours and a more intuitive and easy way to envisage the relationships between the process instances found within the available event data logs and the discovered process models.