Discovery and enhancement of learning model analysis through semantic process mining

International audienceSemantic concepts can be layered on top of existing learner information asset to provide a more conceptual analysis of real time processes capable of providing real world answers that are closer to human understanding. Challenges from current research shows that even though learning data are captured and modelled with acceptable performance to accurately reflect process executions, they are still limited for many process mining analysis because they lack the abstraction level required from real world perspectives. The work in this paper describes a Semantic Process Mining approach directed towards enriching streams of event data logs from a learning process using semantic descriptions that references concepts in an Ontology specifically designed for representing learning processes. The proposed approach involves the extraction of process history data from learning execution environments unfolding how we extract the input data necessary to be mapped unto the learning process logs, which is then followed by submitting the resulting eXtensible Event Streams - XES and Mining eXtensible Markup Language - MXML format to the process analytics environment for mining and further analysis. The consequence is a learning process model which we semantically annotate with concepts they represent in real time using semantic descriptions, and then linking them to an ontology to allow for analysis of the extracted event logs streams based on concepts rather than the event tags of the process. The aim is to provide real time knowledge about the learning process which are more intuitive and closer to human understanding. By referring to ontologies and piloting series of validation experiments, the approach provides us with the capability to infer new and discover relationships the process instances share amongst themselves and to address the problem of determining the presence of different learning patterns within the learning knowledge base. To this end, we demonstrate how data from learning process can be extracted, semantically prepared, and transformed into mining executable formats to enable prediction of individual learning patterns and outcomes through further semantic analysis of the discovered models. Therefore, our approach is grounded on Process Mining and Semantic Modelling Techniques